In
https://github.com/google/tf-quant-finance/tree/master/tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/interest_rate_swap/interest_rate_swap_impl.py

under def ir_delta it says
raise NotImplementedError("Coming soon.")

In ir_delta_parallel_leg, I see that:
if shock_size is None:
bump = tf.constant(0, dtype=self._dtype,
name="bump")
return tff_math.fwd_gradient(price_fn, bump)

So when no shocks are supplied - its just fwd_gradient. Isn't that the AAD equiv of the bump and reval change in PV?

But when a shock size is supplied, you bump the base curve (as a parallel bump) and then:
price_no_bump = leg.price(processed_market_data)
price_with_bump = price_fn(shock_size)
delta = (price_with_bump - price_no_bump) / shock_size
return delta

So in effect, ir_delta_parallel has the potential to supply both delta's? (AAD equiv and parallel shocked

I was wondering if there's an example implementation of BS implied vol estimation for American options and greeks. This tends to be the more commonly needed example than pricing, since the immediate need is to generate vol surfaces or ivol and use in further modelling. Wondering if this is supported, I couldn't see any examples in the American pricing notebook. Thanks!

I haven't found an example of how to properly calculate the greeks with the PDE pricer or the MC pricer.
Here is my attempt : https://colab.research.google.com/github/arthurpham/google_colab/blob/1f737238f1ba71c8c84c47bb24f55e3a97688d1f/AmericanOption_PDE_Greeks_TQF.ipynb
I checked that the greeks with PDE was close to the BS greeks (BS closed form formula and AD).
But the performance of the greeks with PDE was very low, i'm not sure why.
Also i get memory exhaustion on my Google Colab when i try to run a batch larger than 10 options.

What am i missing ?
Thank you

TQF GPU 10 options
wall time: 26.640211820602417
options per second: 0.3753723907054831

Hello,
the following setup returns negative prices for european barrier calls:

import tf_quant_finance as tff
import numpy as np

n_points=10
spots = np.linspace(0.01, 0.2, n_points)
strikes = 1.9 * np.ones(n_points)
volatilities = 0.25 * np.ones(n_points)
expiries = 0.5 * np.ones(n_points)
barriers = 0.5 * np.ones(n_points)
is_barrier_down = np.array([True] * n_points)
is_knock_out = np.array([True] * n_points)
is_call_options = np.array([True] * n_points)

args = {
    "volatilities" : volatilities,
    "strikes" : strikes,
    "expiries" : expiries,
    "spots" : spots,
    "barriers" : barriers,
    "is_barrier_down" : is_barrier_down,
    "is_knock_out" : is_knock_out,
    "is_call_options" : is_call_options,
}

    
price = tff.black_scholes.barrier_price(
        **args
    )

>> price
>> <tf.Tensor: shape=(10,), dtype=float64, numpy=
array([-4.62000000e-01, -3.81777778e-01, -3.01555556e-01, -2.21341484e-01,
       -1.41989785e-01, -7.20212532e-02, -2.70513904e-02, -7.63614881e-03,
       -1.71392373e-03, -3.24709866e-04])>

The parameters are obviously contrived, but i wonder whether this is expected behavior in such tails due to numerical instabilities.

I generated sample paths of HestonModel with parameters shown in the documentation.

However, variance eventually converges to zero and accordingly spot prices asymptotic to constant.
I attach the plot and a snippet to reproduce it.

Is this an expected behaviour?
I appreciate it if you could take a look at it.

import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
import tf_quant_finance as tff
from tf_quant_finance.math import random
from tf_quant_finance.math.piecewise import PiecewiseConstantFunc

tf.random.set_seed(42)

epsilon = PiecewiseConstantFunc(jump_locations=[0.5], values=[1, 1.1], dtype=np.float64)
process = tff.models.HestonModel(
    kappa=0.5, theta=0.04, epsilon=epsilon, rho=0.1, dtype=np.float64
)
times = np.linspace(0.0, 1.0, 1000)
num_samples = 10  # number of trajectories
sample_paths = process.sample_paths(
    times,
    time_step=0.01,
    num_samples=num_samples,
    initial_state=np.array([1.0, 0.04]),
    random_type=random.RandomType.SOBOL,
)

s = np.array(sample_paths[..., 0]).T
v = np.array(sample_paths[..., 1]).T

plt.figure(figsize=(24, 6))
plt.plot(s)
plt.title("spot")
plt.show()

plt.figure(figsize=(24, 6))
plt.plot(v)
plt.title("variance")
plt.show()

$ pip show tf-quant-finance
Name: tf-quant-finance
Version: 0.0.1.dev24
Summary: High-performance TensorFlow library for quantitative finance.
Home-page: https://github.com/google/tf-quant-finance
Author: Google Inc.
Author-email: [email protected]
License: Apache 2.0
Location: /usr/local/lib/python3.8/site-packages
Requires: numpy, protobuf, tensorflow-probability, attrs
Required-by: 

Hi tff team,

Thank you so much for developing this project. My question may be vague. Which precision should I use when doing derivative pricing? Is there any industry standards? Is float32 enough for most cases? Or it depends case by case?

Other quant libraries on CPU, for example QuantLib and QuantLib.jl, they use float64 for almost all scenarios (analytically method, Monte-Carlo or PDE). CPUs usually have float64 support.

I find TFF usually uses tf.float64 when choosing dtype. TFF can run on accelerators: GPUs or TPUs. NVIDIA GPUs have float64 CUDA cores and are capable of handing float64. However, TPUs may not have float64 support. But I find a paper from Google about doing Monte-Carlo on TPUs. They say in most cases, TPUs can do Monte-Carlo. On the hardware side, lower precision usually means faster computing speed. Other academic papers do some experiments on mix-precision.

Is there a simple or quick guide on precision? Like:

• float32 is enough for most Monte-Carlo cases and black-sholes analytically method?
• float64 is required for PDE?

If precision is depended by cases. How to measure the precision is enough for my case?

Thanks!

I did try run from tf_quant_finance directory as well, am I missing something here, surely seems like a path/ pythonpath issue.
Did see in bazel that they remove the sys.path[0], not sure if that's the reason. https://github.com/bazelbuild/bazel/blob/master/tools/python/python_bootstrap_template.txt#L19

Tried to add it to the PYTHONPATH variable as well, but didn't have any luck.

C:\GitRepos\tf-quant-finance (master -> origin)
(tqf) λ bazel test //tf_quant_finance/math/random_ops/sobol:sobol_test
INFO: Analyzed target //tf_quant_finance/math/random_ops/sobol:sobol_test (1 packages loaded, 2 targets configured).
INFO: Found 1 test target...
FAIL: //tf_quant_finance/math/random_ops/sobol:sobol_test (see C:/users/ch/_bazel_ch/ssh2zpbu/execroot/tf_quant_finance/bazel-out/x64_windows-fastbuild/testlogs/tf_quant_finance/math/random_ops/sobol/sobol_test/test.log)
Target //tf_quant_finance/math/random_ops/sobol:sobol_test up-to-date:
  bazel-bin/tf_quant_finance/math/random_ops/sobol/sobol_test.exe
  bazel-bin/tf_quant_finance/math/random_ops/sobol/sobol_test.zip
INFO: Elapsed time: 7.009s, Critical Path: 6.66s
INFO: 2 processes: 2 local.
INFO: Build completed, 1 test FAILED, 2 total actions
//tf_quant_finance/math/random_ops/sobol:sobol_test                      FAILED in 6.6s
  C:/users/ch/_bazel_ch/ssh2zpbu/execroot/tf_quant_finance/bazel-out/x64_windows-fastbuild/testlogs/tf_quant_finance/math/random_ops/sobol/sobol_test/test.log

Executed 1 out of 1 test: 1 fails locally.

Contains of the log file

-----------------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\ch\AppData\Local\Temp\Bazel.runfiles_ue9zden8\runfiles\tf_quant_finance\tf_quant_finance\math\random_ops\sobol\sobol_test.py", line 22, in <module>
    from tf_quant_finance.math import random
ModuleNotFoundError: No module named 'tf_quant_finance.math'

Hi team,
Is there any plan for the implementation of asian option computation? I would like to contribute if there is anything related in you plan.

I can also contribute to other issues if you have one.

Xiaozhi

My goal is to compute american option prices with discrete dividends on a specific date. The example provided for American options does not indicate how could this be done.

# Define the coordinate grid
  s_min = 0.01
  s_max = 300.
  grid = pde.grids.uniform_grid(minimums=[s_min],
                                maximums=[s_max],
                                sizes=[number_grid_points],
                                dtype=dtype)

  # Define the values grid for the final condition
  s = grid[0]
  final_values_grid = tf.nn.relu(s - strike)

  # Define the PDE coefficient functions
  def second_order_coeff_fn(t, grid):
    del t
    s = grid[0]
    return [[volatility ** 2 * s ** 2 / 2]]

  def first_order_coeff_fn(t, grid):
    del t
    s = grid[0]
    return [risk_free_rate * s]

  def zeroth_order_coeff_fn(t, grid):
    del t, grid
    return -risk_free_rate

  # Define the boundary conditions
  @pde.boundary_conditions.dirichlet
  def lower_boundary_fn(t, grid):
    del t, grid
    return tf.constant(0.0, dtype=dtype)

  @pde.boundary_conditions.dirichlet
  def upper_boundary_fn(t, grid):
    del grid
    return tf.squeeze(s_max - strike * tf.exp(-risk_free_rate * (expiry - t)))

  # In order to price American option one needs to set option values to 
  # V(x) := max(V(x), max(x - strike, 0)) after each iteration
  def values_transform_fn(t, grid, values):
    del t
    s = grid[0]
    values_floor = tf.nn.relu(s - strike)
    return grid, tf.maximum(values, values_floor)

Given that discrete options happen at a specific date I thought it would be possible to change the value of spot (s) by the respective dividend after the ex_date occurs. Lets say that expiry = 0.8 and for example ex_date = 0.5, with a dividend of 1.5%, then the code would become:

# Define the coordinate grid
  s_min = 0.01
  s_max = 300.
  grid = pde.grids.uniform_grid(minimums=[s_min],
                                maximums=[s_max],
                                sizes=[number_grid_points],
                                dtype=dtype)

  # Define the values grid for the final condition
  s = grid[0]
  final_values_grid = tf.nn.relu(s - strike)

  # Define the PDE coefficient functions
  def second_order_coeff_fn(t, grid):
    s = grid[0]
    if t.numpy() > 0.5:
        s = s*(1 - 0.015)
    del t
    return [[volatility ** 2 * s ** 2 / 2]]

  def first_order_coeff_fn(t, grid):
    s = grid[0]
    if t.numpy() > 0.5:
        s = s*(1 - 0.015)
    del t
    return [risk_free_rate * s]

  def zeroth_order_coeff_fn(t, grid):
    del t, grid
    return -risk_free_rate

  # Define the boundary conditions
  @pde.boundary_conditions.dirichlet
  def lower_boundary_fn(t, grid):
    del t, grid
    return tf.constant(0.0, dtype=dtype)

  @pde.boundary_conditions.dirichlet
  def upper_boundary_fn(t, grid):
    del grid
    return tf.squeeze(s_max - strike * tf.exp(-risk_free_rate * (expiry - t)))

  # In order to price American option one needs to set option values to 
  # V(x) := max(V(x), max(x - strike, 0)) after each iteration
  def values_transform_fn(t, grid, values):
    s = grid[0]
    if t.numpy() > 0.5:
        s = s*(1 - 0.015)
    del t
    values_floor = tf.nn.relu(s - strike)
    return grid, tf.maximum(values, values_floor)

Unfortunately, this approach does not yield proper results when compared with Quantlib. Does the tff library allow for the computation of option prices using discrete dividends? Ideally the only thing that it is required is to shift all spot prices based on the dividend starting at ex date.

Thank you in advance

I'm observing an issue when trying to calibrate the hull white model using calibration_from_swaptions function. Specifically, when I try to recalculate the prices using the calibrated parameters, I do not return to my source prices. I've included a demo of the issue below:

import numpy as np
import tensorflow as tf
import tf_quant_finance as tff

notional = 100.
prices = np.array([28.54, 27.63, 26.82, 25.76, 24.26])
expiries = np.array([10, 10, 10, 10, 15])
zero_rate_fn = lambda x: 0.01 * tf.ones_like(x, dtype=tf.float64)
start_times = np.array([
    [10. , 10.5, 11. , 11. , 11. , 11. , 11. , 11. , 11. , 11. ],
    [10. , 10.5, 11. , 11.5, 12. , 12. , 12. , 12. , 12. , 12. ],
    [10. , 10.5, 11. , 11.5, 12. , 12.5, 13. , 13. , 13. , 13. ],
    [10. , 10.5, 11. , 11.5, 12. , 12.5, 13. , 13.5, 14. , 14.5],
    [15. , 15.5, 16. , 16. , 16. , 16. , 16. , 16. , 16. , 16. ]])

end_times = np.array([
    [10.5, 11. , 11. , 11. , 11. , 11. , 11. , 11. , 11. , 11. ],
    [10.5, 11. , 11.5, 12. , 12. , 12. , 12. , 12. , 12. , 12. ],
    [10.5, 11. , 11.5, 12. , 12.5, 13. , 13. , 13. , 13. , 13. ],
    [10.5, 11. , 11.5, 12. , 12.5, 13. , 13.5, 14. , 14.5, 15. ],
    [15.5, 16. , 16. , 16. , 16. , 16. , 16. , 16. , 16. , 16. ]])

coupons = np.array([
    [0.03057629, 0.03057629, 0.        , 0.        , 0.        ,
    0.        , 0.        , 0.        , 0.        , 0.        ],
    [0.030797  , 0.030797  , 0.030797  , 0.030797  , 0.        ,
     0.        , 0.        , 0.        , 0.        , 0.        ],
    [0.03072953, 0.03072953, 0.03072953, 0.03072953, 0.03072953,
    0.03072953, 0.        , 0.        , 0.        , 0.        ],
    [0.03086061, 0.03086061, 0.03086061, 0.03086061, 0.03086061,
    0.03086061, 0.03086061, 0.03086061, 0.03086061, 0.03086061],
    [0.02808885, 0.02808885, 0.        , 0.        , 0.        ,
    0.        , 0.        , 0.        , 0.        , 0.        ]])
 
init_mr_guess = [0.05]
init_vol_guess = [0.05]

calibrated_params, converged, _ = tff.models.hull_white.calibration_from_swaptions(
    prices=prices,
    expiries=expiries,
    floating_leg_start_times=start_times,
    floating_leg_end_times=end_times,
    fixed_leg_payment_times=end_times,
    floating_leg_daycount_fractions=end_times - start_times,
    fixed_leg_daycount_fractions=end_times - start_times,
    fixed_leg_coupon=coupons,
    reference_rate_fn=zero_rate_fn,
    mean_reversion=init_mr_guess,
    volatility=init_vol_guess,
    maximum_iterations=50,
    notional=notional,
    use_analytic_pricing=True,
    dtype=tf.float64,
)

print("Calibrated params")
print("Vol:", calibrated_params.volatility.values())
print("Mean reversion:", calibrated_params.mean_reversion.values())
print("Is converged:", converged)
print("")`

recalced_prices = tff.models.hull_white.swaption_price(
    expiries=expiries,
    floating_leg_start_times=start_times,
    floating_leg_end_times=end_times,
    fixed_leg_payment_times=end_times,
    floating_leg_daycount_fractions=end_times-start_times,
    fixed_leg_daycount_fractions=end_times-start_times,
    fixed_leg_coupon=coupons,
    reference_rate_fn=zero_rate_fn,
    notional=notional,
    mean_reversion=calibrated_params.mean_reversion,
    volatility=calibrated_params.volatility,
    dtype=tf.float64,
)

print("Source prices: ", prices)
print("Recalculated prices: ", recalced_prices.numpy())

Output I'm seeing:
Source prices: [28.54, 27.63, 26.82, 25.76, 24.26]
Recalculated prices: [10.43644126, 20.46670051, 29.87954086, 46.01337643, 12.38118779]

Thanks in advance!

Maybe this should be relatively straightforward within tensorflow too.

https://juliacomputing.com/blog/2017/04/26/Wilmott-AD-Greeks.html

AttributeError: module 'QuantLib' has no attribute 'FDAmericanEngine'

According to the quantlib release https://github.com/lballabio/QuantLib/releases/tag/QuantLib-v1.22, FDAmericanEngine has been removed after being deprecated, and is replaced by FdBlackScholesVanillaEngine

Hi
I have successfully installed tensorflow and tf-quant-finance from source, however, when I want to run an example, I get an error about version mismatch of numpy.

$ python3 American_Option_Black_Scholes.py
2021-04-02 08:52:17.384836: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0
RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd
Traceback (most recent call last):
  File "American_Option_Black_Scholes.py", line 8, in <module>
    import tensorflow as tf
  File "/home/mnaderan/.local/lib/python3.8/site-packages/tensorflow/__init__.py", line 41, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "/home/mnaderan/.local/lib/python3.8/site-packages/tensorflow/python/__init__.py", line 40, in <module>
    from tensorflow.python.eager import context
  File "/home/mnaderan/.local/lib/python3.8/site-packages/tensorflow/python/eager/context.py", line 37, in <module>
    from tensorflow.python.client import pywrap_tf_session
  File "/home/mnaderan/.local/lib/python3.8/site-packages/tensorflow/python/client/pywrap_tf_session.py", line 23, in <module>
    from tensorflow.python._pywrap_tf_session import *
ImportError: SystemError: <built-in method __contains__ of dict object at 0x7f4c43c1d600> returned a result with an error set

The versions are shown below

$ pip3 list | grep numpy
numpy                   1.19.5
$ pip3 list | grep tensor
tensorboard             2.4.1
tensorboard-plugin-wit  1.8.0
tensorflow              2.6.0
tensorflow-estimator    2.4.0
tensorflow-probability  0.12.1
$ pip3 list | grep quant
tf-quant-finance        0.0.1.dev25

How can I fix that?

Spread-options are particularly popular in commodity markets. A simple Kirk's approximation for European spread-option price under Black-Scholes model is of interest.

The module implementing this method should live under tf_quant_finance/volatility/spread_option.py. It should support both puts and calls. Tests should be in spread_option_test.py in the same folder.

Team, is there a plan to implement MonteCarlo simulation? I can contribute to one if it is in the works or contribute to any other higher priority items on your list.

Krishna

When we specify the SDE with dimension greater than one and simulate under Euler Scheme, we will have the following errors.

~/anaconda/envs/py36dev/lib/python3.6/site-packages/tf_quant_finance/models/ito_process.py in step_fn(i, written_count, current_state, result)
332 dw = random.mv_normal_sample(
333 (num_samples,), mean=wiener_mean, random_type=random_type,
--> 334 seed=seed)
335 dw = dw * sqrt_dt[i]
336 dt_inc = dt[i] * self.drift_fn()(current_time, current_state) # pylint: disable=not-callable

~/anaconda/envs/py36dev/lib/python3.6/site-packages/tf_quant_finance/math/random_ops/multivariate_normal.py in multivariate_normal(sample_shape, mean, covariance_matrix, scale_matrix, random_type, validate_args, seed, dtype, name, **kwargs)
170 name,
171 default_name='multivariate_normal',
--> 172 values=[sample_shape, mean, covariance_matrix, scale_matrix]):
173 if mean is not None:
174 mean = tf.convert_to_tensor(mean, dtype=dtype, name='mean')

TypeError: init() got an unexpected keyword argument 'default_name'

I'm using Tensorflow 2.0

Hi,

I have just cloned this project and find that Month.md and month.md in api_docs/tf_quant_finance/datetime/ collide on my macOS. I could not pull recent merges. The best solution is to reformat my local APFS filesystem. But I think reformatting filesystem seems not a good choice. I guess api_docs may be generated automatically. However, it would be nice if the api and the markdown file could change to another name. And it would benefit other developers who want to clone the project into their macOS filesystem.

Thanks!

Hi,

I noticed a problem with the time grid in euler_sample when XLA compilation (with dynamic shape for times) is used.

https://colab.research.google.com/drive/1kg8RChmJ3TxdONBHE1lIdCAmZKIXwGcY#scrollTo=pohADCcoTZMh

Thank you

Hello,
The lastest pip package was publish on the 13th of April.
Is there any plan to publish a new version of the lib?

[jw@cn06 tf-quant-finance]$ ./bazel-bin/build_pip_pkg artifacts
++ uname -s
++ tr A-Z a-z

• PLATFORM=linux
• PIP_FILE_PREFIX=bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/
• main artifacts
• SETUP_FLAGS=--universal
• [[ ! -z artifacts ]]
• [[ artifacts == \m\a\k\e ]]
• [[ artifacts == --\n\i\g\h\t\l\y ]]
• DEST=artifacts
• shift
• [[ ! -z '' ]]
• [[ -z artifacts ]]
• mkdir -p artifacts
  ++ readlink -f artifacts
• DEST=/data/jw/tf-quant-finance/artifacts
• echo '=== destination directory: /data/jw/tf-quant-finance/artifacts'
  === destination directory: /data/jw/tf-quant-finance/artifacts
  ++ mktemp -d -t tmp.XXXXXXXXXX
• TMPDIR=/tmp/tmp.IiaeTXyhb0
  ++ date
• echo Fri Mar 5 17:07:50 CET 2021 : '=== Using tmpdir: /tmp/tmp.IiaeTXyhb0'
  Fri Mar 5 17:07:50 CET 2021 : === Using tmpdir: /tmp/tmp.IiaeTXyhb0
• echo '=== Copy TF Quant Finance files'
  === Copy TF Quant Finance files
• cp bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/setup.py /tmp/tmp.IiaeTXyhb0
• cp bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/MANIFEST.in /tmp/tmp.IiaeTXyhb0
• cp bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/LICENSE /tmp/tmp.IiaeTXyhb0
• cp bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/README.md /tmp/tmp.IiaeTXyhb0
• rsync -avm -L '--exclude=*_test.py' bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/tf_quant_finance /tmp/tmp.IiaeTXyhb0
  building file list ... done
  tf_quant_finance/
  tf_quant_finance/init.py
  tf_quant_finance/black_scholes/
  tf_quant_finance/black_scholes/init.py
  tf_quant_finance/black_scholes/brownian_bridge.py
  tf_quant_finance/black_scholes/crr_binomial_tree.py
  tf_quant_finance/black_scholes/implied_vol_approximation.py
  tf_quant_finance/black_scholes/implied_vol_lib.py
  tf_quant_finance/black_scholes/implied_vol_newton_root.py
  tf_quant_finance/black_scholes/implied_vol_utils.py
  tf_quant_finance/black_scholes/vanilla_prices.py
  tf_quant_finance/black_scholes/approximations/
  tf_quant_finance/black_scholes/approximations/init.py
  tf_quant_finance/black_scholes/approximations/american_option.py
  tf_quant_finance/datetime/
  tf_quant_finance/datetime/init.py
  tf_quant_finance/datetime/bounded_holiday_calendar.py
  tf_quant_finance/datetime/constants.py
  tf_quant_finance/datetime/date_tensor.py
  tf_quant_finance/datetime/date_utils.py
  tf_quant_finance/datetime/daycounts.py
  tf_quant_finance/datetime/holiday_calendar.py
  tf_quant_finance/datetime/holiday_calendar_factory.py
  tf_quant_finance/datetime/holiday_utils.py
  tf_quant_finance/datetime/periods.py
  tf_quant_finance/datetime/schedules.py
  tf_quant_finance/datetime/tensor_wrapper.py
  tf_quant_finance/datetime/unbounded_holiday_calendar.py
  tf_quant_finance/experimental/
  tf_quant_finance/experimental/init.py
  tf_quant_finance/experimental/io.py
  tf_quant_finance/experimental/finite_difference/
  tf_quant_finance/experimental/finite_difference/init.py
  tf_quant_finance/experimental/finite_difference/methods.py
  tf_quant_finance/experimental/instruments/
  tf_quant_finance/experimental/instruments/init.py
  tf_quant_finance/experimental/instruments/bond.py
  tf_quant_finance/experimental/instruments/cap_floor.py
  tf_quant_finance/experimental/instruments/cashflow_stream.py
  tf_quant_finance/experimental/instruments/cms_swap.py
  tf_quant_finance/experimental/instruments/eurodollar_futures.py
  tf_quant_finance/experimental/instruments/floating_rate_note.py
  tf_quant_finance/experimental/instruments/forward_rate_agreement.py
  tf_quant_finance/experimental/instruments/interest_rate_swap.py
  tf_quant_finance/experimental/instruments/overnight_index_linked_futures.py
  tf_quant_finance/experimental/instruments/rate_curve.py
  tf_quant_finance/experimental/instruments/rates_common.py
  tf_quant_finance/experimental/instruments/swaption.py
  tf_quant_finance/experimental/local_stochastic_volatility/
  tf_quant_finance/experimental/local_stochastic_volatility/init.py
  tf_quant_finance/experimental/local_stochastic_volatility/local_stochastic_volatility_model.py
  tf_quant_finance/experimental/local_stochastic_volatility/lsv_variance_model.py
  tf_quant_finance/experimental/local_volatility/
  tf_quant_finance/experimental/local_volatility/init.py
  tf_quant_finance/experimental/local_volatility/local_volatility_model.py
  tf_quant_finance/experimental/lsm_algorithm/
  tf_quant_finance/experimental/lsm_algorithm/init.py
  tf_quant_finance/experimental/lsm_algorithm/lsm.py
  tf_quant_finance/experimental/lsm_algorithm/lsm_v2.py
  tf_quant_finance/experimental/lsm_algorithm/payoff.py
  tf_quant_finance/experimental/pricing_platform/
  tf_quant_finance/experimental/pricing_platform/init.py
  tf_quant_finance/experimental/pricing_platform/framework/
  tf_quant_finance/experimental/pricing_platform/framework/init.py
  tf_quant_finance/experimental/pricing_platform/framework/utils.py
  tf_quant_finance/experimental/pricing_platform/framework/core/
  tf_quant_finance/experimental/pricing_platform/framework/core/init.py
  tf_quant_finance/experimental/pricing_platform/framework/core/business_days.py
  tf_quant_finance/experimental/pricing_platform/framework/core/currencies.py
  tf_quant_finance/experimental/pricing_platform/framework/core/curve_types.py
  tf_quant_finance/experimental/pricing_platform/framework/core/daycount_conventions.py
  tf_quant_finance/experimental/pricing_platform/framework/core/implied_volatility_type.py
  tf_quant_finance/experimental/pricing_platform/framework/core/instrument.py
  tf_quant_finance/experimental/pricing_platform/framework/core/interpolation_method.py
  tf_quant_finance/experimental/pricing_platform/framework/core/models.py
  tf_quant_finance/experimental/pricing_platform/framework/core/processed_market_data.py
  tf_quant_finance/experimental/pricing_platform/framework/core/rate_indices.py
  tf_quant_finance/experimental/pricing_platform/framework/core/types.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/init.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/utils.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/american_option/
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/american_option/init.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/american_option/american_option_impl.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/american_option/proto_utils.py
  tf_quant_finance/experimental/pricing_platform/framework/equity_instruments/american_option/utils.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/
  tf_quant_finance/experimental/pricing_platform/framework/market_data/init.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/market_data_config.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/market_data_impl.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/rate_curve.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/utils.py
  tf_quant_finance/experimental/pricing_platform/framework/market_data/volatility_surface.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/init.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/cashflow_streams.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/coupon_specs.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/utils.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/forward_rate_agreement/
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/forward_rate_agreement/init.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/forward_rate_agreement/forward_rate_agreement_impl.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/forward_rate_agreement/proto_utils.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/interest_rate_swap/
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/interest_rate_swap/init.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/interest_rate_swap/interest_rate_swap_impl.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/interest_rate_swap/proto_utils.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/swaption/
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/swaption/init.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/swaption/proto_utils.py
  tf_quant_finance/experimental/pricing_platform/framework/rate_instruments/swaption/swaption_impl.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/
  tf_quant_finance/experimental/pricing_platform/instrument_protos/init.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/all_instruments_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/american_equity_option_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/business_days_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/currencies_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/date_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/daycount_conventions_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/decimal_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/forward_rate_agreement_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/interest_rate_swap_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/metadata_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/period_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/rate_indices_pb2.py
  tf_quant_finance/experimental/pricing_platform/instrument_protos/swaption_pb2.py
  tf_quant_finance/math/
  tf_quant_finance/math/init.py
  tf_quant_finance/math/custom_loops.py
  tf_quant_finance/math/diff_ops.py
  tf_quant_finance/math/gradient.py
  tf_quant_finance/math/jacobian.py
  tf_quant_finance/math/pad.py
  tf_quant_finance/math/piecewise.py
  tf_quant_finance/math/segment_ops.py
  tf_quant_finance/math/integration/
  tf_quant_finance/math/integration/init.py
  tf_quant_finance/math/integration/integrate.py
  tf_quant_finance/math/integration/simpson.py
  tf_quant_finance/math/interpolation/
  tf_quant_finance/math/interpolation/init.py
  tf_quant_finance/math/interpolation/utils.py
  tf_quant_finance/math/interpolation/cubic/
  tf_quant_finance/math/interpolation/cubic/init.py
  tf_quant_finance/math/interpolation/cubic/cubic_interpolation.py
  tf_quant_finance/math/interpolation/interpolation_2d/
  tf_quant_finance/math/interpolation/interpolation_2d/init.py
  tf_quant_finance/math/interpolation/interpolation_2d/interpolation_2d.py
  tf_quant_finance/math/interpolation/linear/
  tf_quant_finance/math/interpolation/linear/init.py
  tf_quant_finance/math/interpolation/linear/linear_interpolation.py
  tf_quant_finance/math/optimizer/
  tf_quant_finance/math/optimizer/init.py
  tf_quant_finance/math/optimizer/conjugate_gradient.py
  tf_quant_finance/math/pde/
  tf_quant_finance/math/pde/init.py
  tf_quant_finance/math/pde/boundary_conditions.py
  tf_quant_finance/math/pde/fd_solvers.py
  tf_quant_finance/math/pde/grids.py
  tf_quant_finance/math/pde/steppers/
  tf_quant_finance/math/pde/steppers/init.py
  tf_quant_finance/math/pde/steppers/composite_stepper.py
  tf_quant_finance/math/pde/steppers/crank_nicolson.py
  tf_quant_finance/math/pde/steppers/douglas_adi.py
  tf_quant_finance/math/pde/steppers/explicit.py
  tf_quant_finance/math/pde/steppers/extrapolation.py
  tf_quant_finance/math/pde/steppers/implicit.py
  tf_quant_finance/math/pde/steppers/multidim_parabolic_equation_stepper.py
  tf_quant_finance/math/pde/steppers/oscillation_damped_crank_nicolson.py
  tf_quant_finance/math/pde/steppers/parabolic_equation_stepper.py
  tf_quant_finance/math/pde/steppers/weighted_implicit_explicit.py
  tf_quant_finance/math/random_ops/
  tf_quant_finance/math/random_ops/init.py
  tf_quant_finance/math/random_ops/multivariate_normal.py
  tf_quant_finance/math/random_ops/stateless.py
  tf_quant_finance/math/random_ops/uniform.py
  tf_quant_finance/math/random_ops/halton/
  tf_quant_finance/math/random_ops/halton/init.py
  tf_quant_finance/math/random_ops/halton/halton_impl.py
  tf_quant_finance/math/random_ops/sobol/
  tf_quant_finance/math/random_ops/sobol/init.py
  tf_quant_finance/math/random_ops/sobol/sobol_impl.py
  tf_quant_finance/math/root_search/
  tf_quant_finance/math/root_search/init.py
  tf_quant_finance/math/root_search/brent.py
  tf_quant_finance/math/root_search/newton.py
  tf_quant_finance/math/root_search/utils.py
  tf_quant_finance/models/
  tf_quant_finance/models/init.py
  tf_quant_finance/models/euler_sampling.py
  tf_quant_finance/models/generic_ito_process.py
  tf_quant_finance/models/ito_process.py
  tf_quant_finance/models/joined_ito_process.py
  tf_quant_finance/models/milstein_sampling.py
  tf_quant_finance/models/utils.py
  tf_quant_finance/models/geometric_brownian_motion/
  tf_quant_finance/models/geometric_brownian_motion/init.py
  tf_quant_finance/models/geometric_brownian_motion/multivariate_geometric_brownian_motion.py
  tf_quant_finance/models/geometric_brownian_motion/univariate_geometric_brownian_motion.py
  tf_quant_finance/models/heston/
  tf_quant_finance/models/heston/init.py
  tf_quant_finance/models/heston/heston_model.py
  tf_quant_finance/models/heston/approximations/
  tf_quant_finance/models/heston/approximations/init.py
  tf_quant_finance/models/heston/approximations/european_option.py
  tf_quant_finance/models/hjm/
  tf_quant_finance/models/hjm/init.py
  tf_quant_finance/models/hjm/cap_floor.py
  tf_quant_finance/models/hjm/quasi_gaussian_hjm.py
  tf_quant_finance/models/hjm/swaption_pricing.py
  tf_quant_finance/models/hjm/swaption_util.py
  tf_quant_finance/models/hjm/zero_coupon_bond_option.py
  tf_quant_finance/models/hjm/zero_coupon_bond_option_util.py
  tf_quant_finance/models/hull_white/
  tf_quant_finance/models/hull_white/init.py
  tf_quant_finance/models/hull_white/calibration.py
  tf_quant_finance/models/hull_white/cap_floor.py
  tf_quant_finance/models/hull_white/one_factor.py
  tf_quant_finance/models/hull_white/swaption.py
  tf_quant_finance/models/hull_white/vector_hull_white.py
  tf_quant_finance/models/hull_white/zero_coupon_bond_option.py
  tf_quant_finance/models/sabr/
  tf_quant_finance/models/sabr/init.py
  tf_quant_finance/models/sabr/sabr_model.py
  tf_quant_finance/rates/
  tf_quant_finance/rates/init.py
  tf_quant_finance/rates/cashflows.py
  tf_quant_finance/rates/forwards.py
  tf_quant_finance/rates/swap_curve_bootstrap.py
  tf_quant_finance/rates/swap_curve_common.py
  tf_quant_finance/rates/swap_curve_fit.py
  tf_quant_finance/rates/analytics/
  tf_quant_finance/rates/analytics/init.py
  tf_quant_finance/rates/analytics/cashflows.py
  tf_quant_finance/rates/analytics/forwards.py
  tf_quant_finance/rates/analytics/swap.py
  tf_quant_finance/rates/constant_fwd/
  tf_quant_finance/rates/constant_fwd/init.py
  tf_quant_finance/rates/constant_fwd/constant_fwd_interpolation.py
  tf_quant_finance/rates/hagan_west/
  tf_quant_finance/rates/hagan_west/init.py
  tf_quant_finance/rates/hagan_west/bond_curve.py
  tf_quant_finance/rates/hagan_west/monotone_convex.py
  tf_quant_finance/utils/
  tf_quant_finance/utils/init.py
  tf_quant_finance/utils/tf_functions.py

sent 1,990,664 bytes received 3,939 bytes 3,989,206.00 bytes/sec
total size is 1,974,560 speedup is 0.99

• rsync -avm -L '--exclude=*_test.py' bazel-bin/build_pip_pkg.runfiles/tf_quant_finance/third_party /tmp/tmp.IiaeTXyhb0/tf_quant_finance
  building file list ... done
  third_party/
  third_party/sobol_data/
  third_party/sobol_data/LICENSE
  third_party/sobol_data/new-joe-kuo-6.21201

sent 1,890,131 bytes received 60 bytes 3,780,382.00 bytes/sec
total size is 1,889,431 speedup is 1.00

• pushd /tmp/tmp.IiaeTXyhb0
  /tmp/tmp.IiaeTXyhb0 /data/jw/tf-quant-finance
  ++ date
• echo Fri Mar 5 17:07:50 CET 2021 : '=== Building wheel'
  Fri Mar 5 17:07:50 CET 2021 : === Building wheel
• python3 setup.py bdist_wheel --universal
  usage: setup.py [global_opts] cmd1 [cmd1_opts] [cmd2 [cmd2_opts] ...]
  or: setup.py --help [cmd1 cmd2 ...]
  or: setup.py --help-commands
  or: setup.py cmd --help

error: invalid command 'bdist_wheel'
[jw@cn06 tf-quant-finance]$

Hi!

Thanks for the really cool library! I recently came across this library as I want to learn more about quant finance and ML and this package really interested me :) I understand it's currently pre-Alpha but I am wondering if this package is compatible with TF 2.0, if not, is it somewhere on the roadmap or is that something I can help out with? Thank you so much!

Trying to run the example Monte_Carlo_Euler_Scheme.ipynb I run into this error:
TypeError: Can't instantiate abstract class ExampleProcess with abstract methods fd_solver_backward, sample_paths

https://colab.research.google.com/github/google/tf-quant-finance/blob/master/tf_quant_finance/examples/jupyter_notebooks/Monte_Carlo_Euler_Scheme.ipynb#scrollTo=uG8UAZXjeUhf

There is some problem on the reported time for GPU vs CPU on multiple dimensions 100.
I am using a single GPU and it is slower compared to my CPU multithreading. I am wondering why... The rest of the notebooks work as expected.
I am using a GeForce 1080 not a TESLA v100 GPU, but intuitively i would expect to have a similar speedup...

Any ideas why I get this behaviour ?

Regards,
pb

Hello and thanks for the library!

I am working on an integration of TFF into one of my option trading strategies and trying to figure out several practical aspects, such as IV determination and finding greeks. After reading through all issues and discussions I have a working notebook which expands the American option example (using PDE) and includes volatility estimation and (most of) greeks using tff.math.fwd_gradient. There are still some questions about gridding to manage batches with significantly different strikes, but I think I'll manage, maybe even to an extent I can do a PR for that example...

One thing I cannot comprehend yet is the time decay and time aspects overalls. How would you recommend to approach theta estimation? Should I try fwd_gradient on expiry? Related question is how to manage options with significantly different time to expiration? The example states that volatility term has to be adjusted - not sure I understand it.

expiry: A Python float. Expiry date of the options. If the options
      have different expiries, volatility term has to adjusted to
      make expiries the same.

I am currently having issues in matching BBG discount OIS curve using the same SOFR swaps used in the icvs function. I suspect this is due to the different compounding of libor an sofr. Is there any workaround to this issue?

I'm trying to replicate the TARF payoff from this paper.
So far, i've been able to write a payoff that can be evaluated with MC and GPU, but i get an error with the XLA compilation.
https://colab.research.google.com/github/arthurpham/google_colab/blob/7d5715c2b0f988c8d7d544fea0ff741e63272056/TARF_MC_Performance_TQF.ipynb
I understand that the payoff code style might not be the best, but i wasn't able to find an example of a path-dependent payoff with MC.
Any pointer would be helpful.

---------------------------------------------------------------------------
InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-7-945d7282d2b1> in <module>()
      1 device = "/gpu:0"
      2 with tf.device(device):
----> 3     tarf_price = price_eu_options_xla(strikes, spot, sigma)
      4     print('price', tarf_price)

1 frames
/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
     53     ctx.ensure_initialized()
     54     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
---> 55                                         inputs, attrs, num_outputs)
     56   except core._NotOkStatusException as e:
     57     if name is not None:

InvalidArgumentError: Detected unsupported operations when trying to compile graph __inference_price_eu_options_9107[_XlaMustCompile=true,config_proto=6001324581131673121,executor_type=11160318154034397263] on XLA_GPU_JIT: TensorListReserve (No registered 'TensorListReserve' OpKernel for XLA_GPU_JIT devices compatible with node {{function_node __inference_while_fn_9086}}{{node while_init/TensorArrayV2_7}}
	 (OpKernel was found, but attributes didn't match) Requested Attributes: element_dtype=DT_VARIANT, shape_type=DT_INT32){{function_node __inference_while_fn_9086}}{{node while_init/TensorArrayV2_7}}
The op is created at: 
File "/usr/lib/python3.7/runpy.py", line 193, in _run_module_as_main
  "__main__", mod_spec)
File "/usr/lib/python3.7/runpy.py", line 85, in _run_code
  exec(code, run_globals)
File "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py", line 16, in <module>
  app.launch_new_instance()
File "/usr/local/lib/python3.7/dist-packages/traitlets/config/application.py", line 846, in launch_instance
  app.start()
File "/usr/local/lib/python3.7/dist-packages/ipykernel/kernelapp.py", line 499, in start
  self.io_loop.start()
File "/usr/local/lib/python3.7/dist-packages/tornado/platform/asyncio.py", line 132, in start
  self.asyncio_loop.run_forever()
File "/usr/lib/python3.7/asyncio/base_events.py", line 541, in run_forever
  self._run_once()
File "/usr/lib/python3.7/asyncio/base_events.py", line 1786, in _run_once
  handle._run()
File "/usr/lib/python3.7/asyncio/events.py", line 88, in _run
  self._context.run(self._callback, *self._args)
File "/usr/local/lib/python3.7/dist-packages/tornado/ioloop.py", line 758, in _run_callback
  ret = callback()
File "/usr/local/lib/python3.7/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
  return fn(*args, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/zmq/eventloop/zmqstream.py", line 536, in <lambda>
  self.io_loop.add_callback(lambda: self._handle_events(self.socket, 0))
File "/usr/local/lib/python3.7/dist-packages/zmq/eventloop/zmqstream.py", line 452, in _handle_events
  self._handle_recv()
File "/usr/local/lib/python3.7/dist-packages/zmq/eventloop/zmqstream.py", line 481, in _handle_recv
  self._run_callback(callback, msg)
File "/usr/local/lib/python3.7/dist-packages/zmq/eventloop/zmqstream.py", line 431, in _run_callback
  callback(*args, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/tornado/stack_context.py", line 300, in null_wrapper
  return fn(*args, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py", line 283, in dispatcher
  return self.dispatch_shell(stream, msg)
File "/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py", line 233, in dispatch_shell
  handler(stream, idents, msg)
File "/usr/local/lib/python3.7/dist-packages/ipykernel/kernelbase.py", line 399, in execute_request
  user_expressions, allow_stdin)
File "/usr/local/lib/python3.7/dist-packages/ipykernel/ipkernel.py", line 208, in do_execute
  res = shell.run_cell(code, store_history=store_history, silent=silent)
File "/usr/local/lib/python3.7/dist-packages/ipykernel/zmqshell.py", line 537, in run_cell
  return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)
File "/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py", line 2718, in run_cell
  interactivity=interactivity, compiler=compiler, result=result)
File "/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py", line 2822, in run_ast_nodes
  if self.run_code(code, result):
File "/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py", line 2882, in run_code
  exec(code_obj, self.user_global_ns, self.user_ns)
File "<ipython-input-7-945d7282d2b1>", line 3, in <module>
  tarf_price = price_eu_options_xla(strikes, spot, sigma)
File "<ipython-input-5-1d20baab715c>", line 89, in price_eu_options
  payoffs = tf.vectorized_map(my_function, reshaped_paths) [Op:__inference_price_eu_options_9107]

Hi guys,

In the "Monte Carlo via Euler Scheme" example you compare TF with QuantLib pricing and conclude that TF finance is x100 times faster(or more).

I want to note that in QL you evolve 100 time steps of Log Normal process, but in TF you work in log space and only apply exp() at the end.
I agree QL may not be very fast, but in this example you compare 100 exponents per path in QL to just 1 exponent in TF...

Thank you!

The following input seems to give bad output with a strange jump,

reference_times = np.array([0.08219178, 0.12328767, 0.16438356, 0.24657534, 0.32876712, 0.4109589, 0.49315068, 0.98630137, 1.02191781], dtype=np.float64)
yields = np.array([2.5e-5, 1.189e-4, 2.151e-4, 4.236e-4, 6.5e-4, 8.905e-4, 1.141e-3, 2.6321e-3, 4.366e-3], dtype=np.float64)
interp_times = np.linspace(0.1, 1.0, 100, dtype=np.float64)
interp_yields = monotone_convex.interpolate_yields(interp_times, reference_times, yields=yields)

Plotting it out gives,

Can someone kindly advise?

Regards,

Are there any plans (or objections) to add this to conda-forge? Given that it's a prue Python library, it should be rather straight-forward to do.

American options have a number of approximate pricing formula under Black-Scholes.
In particular, the Baron-Adesi Whaley approximation is of interest.

The module implementing this method should live under tf_quant_finance/volatility/american_option.py. It should support both puts and calls. Tests should be in american_option_test.py in the same folder.

I am using tf-quant-finance for plotting option values and greeks. When plotting at-expiration (so expiries=0) most of the models return nan for some or all prices. Is this intended behavior, in which case I should wrap my calls to tf-quant-finance with some code to counter this, or is this something that could be changed?

This issue shows up for:

• Bjerksund-Stensland with modified_boundary=True for all prices on a call, but only out-of-the-money prices on puts
• Adesi-Whaley for out-of-the-money prices on calls only
• Binomial for all options (American or European, calls or puts)

Only Black-Scholes-Merton and Bjerksund-Stensland with modified_boundary=False are not affected by this.

I could submit a PR to fix this issue, but before I do the work I want to make sure that this is a bug and not a feature.

Heston model has accurate density approximations for European option prices, which are of interest.

The module implementing this method should live under tf_quant_finance/volatility/heston_approximation.py. It should support both European option puts and calls approximations. Tests should be in heston_approximation_test.py in the same folder.

I installed tf-quant-finance from source cloned on 16-Jun-2021 and get the below error:

>>> import tf_quant_finance as tff
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/tf-quant-finance/tf_quant_finance/__init__.py", line 74, in <module>
    from tf_quant_finance import experimental
  File "/tf-quant-finance/tf_quant_finance/experimental/__init__.py", line 19, in <module>
    from tf_quant_finance.experimental import local_stochastic_volatility
  File "/tf-quant-finance/tf_quant_finance/experimental/local_stochastic_volatility/__init__.py", line 17, in <module>
    from tf_quant_finance.experimental.local_stochastic_volatility.local_stochastic_volatility_model import LocalStochasticVolatilityModel
  File "/tf-quant-finance/tf_quant_finance/experimental/local_stochastic_volatility/local_stochastic_volatility_model.py", line 25, in <module>
    from tf_quant_finance.experimental import local_volatility as lvm
  File "/tf-quant-finance/tf_quant_finance/experimental/local_volatility/__init__.py", line 18, in <module>
    from tf_quant_finance.experimental.local_volatility.local_volatility_model import LocalVolatilityModel
  File "/tf-quant-finance/tf_quant_finance/experimental/local_volatility/local_volatility_model.py", line 25, in <module>
    from tf_quant_finance.experimental.pricing_platform.framework.market_data import volatility_surface
  File "/tf-quant-finance/tf_quant_finance/experimental/pricing_platform/__init__.py", line 17, in <module>
    from tf_quant_finance.experimental.pricing_platform import framework
  File "/tf-quant-finance/tf_quant_finance/experimental/pricing_platform/framework/__init__.py", line 17, in <module>
    from tf_quant_finance.experimental.pricing_platform.framework import core
  File "/tf-quant-finance/tf_quant_finance/experimental/pricing_platform/framework/core/__init__.py", line 17, in <module>
    from tf_quant_finance.experimental.pricing_platform.framework.core import business_days
  File "/tf-quant-finance/tf_quant_finance/experimental/pricing_platform/framework/core/business_days.py", line 18, in <module>
    from tf_quant_finance.experimental.pricing_platform.instrument_protos import business_days_pb2
  File "/tf-quant-finance/tf_quant_finance/experimental/pricing_platform/instrument_protos/__init__.py", line 17, in <module>
    from tf_quant_finance.experimental.pricing_platform.instrument_protos import all_instruments_pb2 as instruments
ImportError: cannot import name 'all_instruments_pb2' from partially initialized module 'tf_quant_finance.experimental.pricing_platform.instrument_protos' (most likely due to a circular import)

This looks to be a bug which was introduced recently (last time I installed from source on 18-May-2021 I did not get this error).

I am not familiar with any of the files mentioned here, so would not be the best person to investigate. Is there someone who can investigate this? Do others get the same error when installing from source?

Formulas for pricing a Barrier option under Black-Scholes model is of interest. (See, e.g., Section 26.9 of Hull(2018), Options, Futures, and Other Derivatives, 9th edition).

The module implementing this method should live under tf_quant_finance/volatility/barrier_option.py. It should support both puts (up-and-in put, down-and-out put) and calls (down-in call, up-and-out call). Tests should be in barrier_option_test.py in the same folder.

Argument names for dividends across different option pricing models appear to be different, with no obvious reason for the difference. For example, tff.black_scholes.option_price takes in continuous_dividends, while tff.black_scholes.option_price_binomial takes in dividend_rates.

If there is no reason for the difference, then I would suggest changing dividend_rates to continuous_dividends, as tff.black_scholes.approximations.adesi_whaley also takes continuous_dividends. This would make it easier to write reusable code independent of the pricing model used.

What do others think of this? If everyone agrees to this change, then I can make it and submit the pull request.

SABR model has accurate density approximations (see, e.g., here). It is of interest to use the approximations to estimate European option prices.

The module implementing this method should live under tf_quant_finance/volatility/sabr_approximation.py. It should support both European puts and calls approximations. Tests should be in sabr_approximation_test.py in the same folder.

Levenberg-Marquardt is a frequently needed algorithm in curve fitting problems in finance. It would be great to have a batched TF implementation in the library. It should be placed in the optimizers folder.

The reference implementation to follow is the MINPACK routines. In addition to the original Fortran version there is also a C++ version available.

Bjerksund-Stensland approximation is an alternative to the Baron-Adesi Whaley approximation for American options pricing formula under Black-Scholes.

The module implementing this method should live under tf_quant_finance/volatility/american_option.py. It should support both puts and calls. Tests should be in american_option_test.py in the same folder.

https://colab.research.google.com/github/arthurpham/google_colab/blob/e20469359e82ecf96640837e2022f58de0df8314/AmericanOption_MC_MemoryLeak_TQF.ipynb

I've tried to reuse the jupyter notebook that show how to do a MC pricing.

for i in trange(1, 100000):
    strikes = tf.constant(np.random.rand(i) * (0.5*spot) + spot, dtype )
    price_eu_options(strikes, spot, sigma)

The test is slightly silly but if the application that calls the pricing library might chose to send an arbitrary strike size, so the batch size might not be known in advance, so i'm wondering what is the best way to handle that without asking quants to micromanage the shapes of each inputs (as a change in the shape of one input retrigger the graph construction) ?

Thank you

Hi!
The main page mentions:
tf-quant-finance: Open mailing list for discussion and questions of this library.
However, the group isn't findable and subscribing via email reveals its settings are "Invite Only".
I would suggest to remedy this by either adjusting the settings or updating the info on the main page to remove the list.
Thank you,

Hello,

I am the chair of a working group at Khronos developing standards for graphics and compute hardware accelerators (as well as a representative for Nvidia in those groups).

We are organizing a summit to gather feedback from the ecosystem, and influence the design of the next round of improvements, and I thought that you may be interested in being represented:
https://www.khronos.org/events/2021-invitation-to-the-khronos-machine-learning-summit

This summit is IP free, and will let you present your project and your needs for improvements in the ML ecosystem, as well as hearing from other companies.

regards,

Pierre Boudier
Software architect at Nvidia
chair of the Machine Learning TSG at Khronos

I was refactoring the "American_Option_Black_Sholes.ipynb" to use the new fd_solvers API with TensorFlow 2.1, but it has much worse performance than in the previous version that was pre-TF2.0.

Here is the notebook re-written in eager style and ran on the colab GPU: https://colab.research.google.com/drive/1FVYloxIrSvlSrp1aMPLxN4f4KlvPI8l0

As you can see the CPU and GPU options/sec benchmarks are both around 1/3 what they were in the old notebook saved on the master branch. I've also tried running the same code, but using the tf.compat.v1 Session and graph; this did not improve performance at all. I've also tried to profile the code using tf.summary.trace_on(graph=True, profiler=True), but it doesn't produce any useful information.

Is this performance something I'm doing wrong with my implementation, is there perhaps some better way I can profile the core library for issues, or is this related to some performance issues with TF2.0 itself?

Binomial options pricing models like Cox-Ross-Rubinstein have been widely used since they are able to handle a variety of conditions for which other models cannot easily be applied proving themselves to be more accurate, particularly for longer-dated options on securities with dividend payments.

Hi all, many thanks for this library! As I've seen, there is an integration method and interpolation. The latter are also available in other packages (tf-graphics, tf-addons) in one way or another while the integration is not (AFAIK?).

However, we implemented multiple integration algorithms and plan to put them up/into an existing repository. TFP seems nice but maybe not fully suited. However, this repository also does not seem directly like the place for this methods (as it is a very general used functionality, especially in science). Or what do you think?

Would it make sense to put this somewhere else? If so, where?

Many thanks for all the work!

pricing and calibration of swpations thanks the LMM with following features

• displacement for negative rates
• number of factors
• stochastic vol
  https://cermics.enpc.fr/~mehallas/Documents/Summer_school_X.pdf

Do you have a running example of tff.math.fwd_gradient either in the notebook or somewhere in the tests, only saw a very contrived one in https://github.com/google/tf-quant-finance/blob/39a1c03cc9938c6e08fceb63e405eff0fda835be/tf_quant_finance/math/integration/integration_test.py

Mainly for the purposes of calculating:
a) first order greeks
b) higher order greeks

Hello,
I'm trying to use the SABR calibration with real BTC option prices but it stops right away without any error message (is_converged is False).
If anyone could help to see what is wrong with this pretty simple code:
Colab notebook (created a wheel with the latest code)

import pandas as pd
import requests
import tf_quant_finance as tff
import time
from datetime import datetime
import numpy

vals = pd.DataFrame(requests.get(
  'https://www.deribit.com/api/v2/public/get_instruments?currency=BTC&expired=false&kind=option').json()['result'])
jul_exp = vals[(vals['expiration_timestamp'] == 1627632000000)]
jul_exp['price'] = [requests.get(f'https://www.deribit.com/api/v2/public/ticker?instrument_name={x}').json()['result']['last_price'] for x in jul_exp['instrument_name']]
jul_exp.dropna(subset=['price'], inplace = True)
underlying = requests.get('https://www.deribit.com/api/v2/public/ticker?instrument_name=BTC-30JUL21').json()['result']
jul_exp['price'] = jul_exp['price'] * underlying['index_price']
is_call = numpy.array([[t == 'call' for t in jul_exp['option_type']]], dtype=numpy.bool)
delta =(( jul_exp.iloc[0]['expiration_timestamp']/1000 - time.time())  / 60 / 60 / 24) / 365
dtype = numpy.float64
models, is_converged, it = tff.models.sabr.approximations.calibration(
    prices=numpy.array([jul_exp['price'].to_numpy()], dtype=dtype),
    strikes=numpy.array([jul_exp['strike'].to_numpy()], dtype=dtype),
    expiries=numpy.array([delta], dtype=dtype),
    forwards=numpy.array([underlying['index_price']], dtype=dtype),
    is_call_options=is_call,
    beta=numpy.array([0.5], dtype=dtype),
    calibrate_beta=True,
    nu=numpy.array([0.1], dtype=dtype),
    nu_lower_bound=0.0,
    nu_upper_bound=10.0,
    rho=numpy.array([0.0], dtype=dtype),
    rho_lower_bound=-0.75,
    rho_upper_bound=0.75,
    maximum_iterations=1000)

when cloning, the following warning pops up:

warning: the following paths have collided (e.g. case-sensitive paths
on a case-insensitive filesystem) and only one from the same
colliding group is in the working tree:

'api_docs/tf_quant_finance/datetime/Month.md'
'api_docs/tf_quant_finance/datetime/month.md'