Skip to content

Commit

Permalink
Optimizer Notebooks
Browse files Browse the repository at this point in the history
  • Loading branch information
@sklbancor
sklbancor committed May 3, 2024
1 parent 1be332a commit 768f75a
Show file tree
Hide file tree
Showing 7 changed files with 2,120 additions and 626 deletions.
50 changes: 25 additions & 25 deletions resources/NBTest/Optimizer_2312_THOR.ipynb
View file

Large diffs are not rendered by default.

516 changes: 369 additions & 147 deletions resources/NBTest/Optimizer_2405_SEI.ipynb
View file

Large diffs are not rendered by default.

130 changes: 70 additions & 60 deletions resources/NBTest/Optimizer_2405_SEI.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,88 +17,65 @@
try:
from fastlane_bot.tools.simplepair import SimplePair
from fastlane_bot.tools.cpc import ConstantProductCurve, CPCContainer
from fastlane_bot.tools.optimizer import PairOptimizer, MargPOptimizer
from fastlane_bot.tools.optimizer import CPCArbOptimizer, PairOptimizer, MargPOptimizer
except:
from tools.simplepair import SimplePair
from tools.cpc import ConstantProductCurve, CPCContainer
from tools.optimizer import PairOptimizer, MargPOptimizer
from tools.optimizer import CPCArbOptimizer, PairOptimizer, MargPOptimizer
CPC = ConstantProductCurve

import pandas as pd

print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(SimplePair))
print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(CPC))
print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(CPCContainer))
print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(CPCArbOptimizer))
print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(PairOptimizer))
print("{0.__name__} v{0.__VERSION__} ({0.__DATE__})".format(MargPOptimizer))
# -

# # Optimizer Testing (SEI, May 2024)

# This is a light workbook allowing to look at issues that may arise when running the optimizer on a specific set of curves.
#
# Instructions:
#
# - locate the **exact** curve set to feed to the optimizer (it will be somewhere in the logging output, and it will be a list of ConstantProductCurve objects)
# - assign it to the `CurvesRaw` variable as shown below
# - add the missing token addresses to the `TOKENS` dict below
# - provide consistent values for `PSTART`
# - run the workbook

# ### >> Enter curves
#
# Place curves here in the form
#
# CurvesRaw = [
# ConstantProductCurve(k=27518385.40998667, x=1272.2926367501436, x_act=0, ...),
# ConstantProductCurve(k=6.160500599566333e+18, x=11099999985.149971, x_act=0, ...),
# ...
# ]

CurvesRaw = [
# `CurvesRaw1` is the original curve set exhibiting the convergence problem

CurvesRaw1 = [
ConstantProductCurve(k=27518385.40998667, x=1272.2926367501436, x_act=0, y_act=2000.9999995236503, alpha=0.5, pair='0x514910771AF9Ca656af840dff83E8264EcF986CA/0x8E870D67F660D95d5be530380D0eC0bd388289E1', cid='0x425d5d4ad7243f88d9f4cde8da52863b45af1f64e05bede1299909bcaa6c52d1-0', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 2000.9999995236503, 'yint': 2000.9999995236503, 'A': 0.38144823884371704, 'B': 3.7416573867739373, 'pa': 16.99999999999995, 'pb': 13.99999999999997}),
ConstantProductCurve(k=6.160500599566333e+18, x=11099999985.149971, x_act=0, y_act=55.50000002646446, alpha=0.5, pair='0x8E870D67F660D95d5be530380D0eC0bd388289E1/0x514910771AF9Ca656af840dff83E8264EcF986CA', cid='0x425d5d4ad7243f88d9f4cde8da52863b45af1f64e05bede1299909bcaa6c52d1-1', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 55.50000002646446, 'yint': 55.50000002646446, 'A': 0, 'B': 0.22360678656963742, 'pa': 0.04999999999999889, 'pb': 0.04999999999999889}),
ConstantProductCurve(k=14449532.299465338, x=57487.82879658422, x_act=0, y_act=5.0, alpha=0.5, pair='0x514910771AF9Ca656af840dff83E8264EcF986CA/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', cid='0x3fcccfe0063b71fc973fab8dea39b6be9da80125910c10e57b924b3e4687295a-0', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 5.0, 'yint': 8.582730309868262, 'A': 0.002257868117407469, 'B': 0.06480740698407672, 'pa': 0.004497751124437756, 'pb': 0.004199999999999756}),
ConstantProductCurve(k=14456757.06563651, x=251.4750925240284, x_act=0, y_act=807.9145301701096, alpha=0.5, pair='0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE/0x514910771AF9Ca656af840dff83E8264EcF986CA', cid='0x3fcccfe0063b71fc973fab8dea39b6be9da80125910c10e57b924b3e4687295a-1', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 807.9145301701096, 'yint': 1974.7090228584536, 'A': 0.519359008452966, 'B': 14.907119849998594, 'pa': 237.97624997025295, 'pb': 222.22222222222211}),
ConstantProductCurve(k=56087178.30932376, x=131.6236694086859, x_act=0, y_act=15920.776548455418, alpha=0.5, pair='0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE/0x8E870D67F660D95d5be530380D0eC0bd388289E1', cid='0x6cc4b198ec4cf17fdced081b5611279be73e200711238068b5340e606ba86646-0', fee=2000, descr='carbon_v1 0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 15920.776548455418, 'yint': 32755.67010983316, 'A': 4.373757425036729, 'B': 54.77225575051648, 'pa': 3498.2508745627138, 'pb': 2999.9999999999854}),
ConstantProductCurve(k=56059148.73497429, x=426117.72306081816, x_act=0, y_act=5.0, alpha=0.5, pair='0x8E870D67F660D95d5be530380D0eC0bd388289E1/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', cid='0x6cc4b198ec4cf17fdced081b5611279be73e200711238068b5340e606ba86646-1', fee=2000, descr='carbon_v1 0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 5.0, 'yint': 10.106093048875099, 'A': 0.0013497708452092638, 'B': 0.016903085094568837, 'pa': 0.0003331667499582927, 'pb': 0.0002857142857142352})
ConstantProductCurve(k=14449532.299465338, x=57487.82879658422, x_act=0, y_act=5.0, alpha=0.5, pair='0x514910771AF9Ca656af840dff83E8264EcF986CA/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2', cid='0x3fcccfe0063b71fc973fab8dea39b6be9da80125910c10e57b924b3e4687295a-0', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 5.0, 'yint': 8.582730309868262, 'A': 0.002257868117407469, 'B': 0.06480740698407672, 'pa': 0.004497751124437756, 'pb': 0.004199999999999756}),
ConstantProductCurve(k=14456757.06563651, x=251.4750925240284, x_act=0, y_act=807.9145301701096, alpha=0.5, pair='0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2/0x514910771AF9Ca656af840dff83E8264EcF986CA', cid='0x3fcccfe0063b71fc973fab8dea39b6be9da80125910c10e57b924b3e4687295a-1', fee=2000, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 807.9145301701096, 'yint': 1974.7090228584536, 'A': 0.519359008452966, 'B': 14.907119849998594, 'pa': 237.97624997025295, 'pb': 222.22222222222211}),
ConstantProductCurve(k=56087178.30932376, x=131.6236694086859, x_act=0, y_act=15920.776548455418, alpha=0.5, pair='0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2/0x8E870D67F660D95d5be530380D0eC0bd388289E1', cid='0x6cc4b198ec4cf17fdced081b5611279be73e200711238068b5340e606ba86646-0', fee=2000, descr='carbon_v1 0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 15920.776548455418, 'yint': 32755.67010983316, 'A': 4.373757425036729, 'B': 54.77225575051648, 'pa': 3498.2508745627138, 'pb': 2999.9999999999854}),
ConstantProductCurve(k=56059148.73497429, x=426117.72306081816, x_act=0, y_act=5.0, alpha=0.5, pair='0x8E870D67F660D95d5be530380D0eC0bd388289E1/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2', cid='0x6cc4b198ec4cf17fdced081b5611279be73e200711238068b5340e606ba86646-1', fee=2000, descr='carbon_v1 0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE\\/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'y': 5.0, 'yint': 10.106093048875099, 'A': 0.0013497708452092638, 'B': 0.016903085094568837, 'pa': 0.0003331667499582927, 'pb': 0.0002857142857142352})
]
CCRaw1 = CPCContainer(CurvesRaw1)

# `CurvesRaw2` is a modified curve set designed to address the problem. On those the algo should converge without issue.

CurvesRaw2 = [
ConstantProductCurve(k=27518385.40998667, x=1272.2926367501436, x_act=0, y_act=2000.9999995236503, alpha=0.5, pair='0x514910771AF9Ca656af840dff83E8264EcF986CA/0x8E870D67F660D95d5be530380D0eC0bd388289E1', cid='0x6673a1e11aed37f08479f05b6f8e492eea344f4b21fa5cdfc2abb58d0a0339f6-0', fee=2000.0, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'tknx_dec': 18, 'tkny_dec': 18, 'tknx_addr': '0x514910771AF9Ca656af840dff83E8264EcF986CA', 'tkny_addr': '0x8E870D67F660D95d5be530380D0eC0bd388289E1', 'blocklud': 19764257, 'y': 2000.9999995236503, 'yint': 2000.9999995236503, 'A': 0.38144823884371704, 'B': 3.7416573867739373, 'pa': 16.99999999999995, 'pb': 13.99999999999997, 'minrw': 1e-06}),
ConstantProductCurve(k=14269669.033092778, x=57121.7742265515, x_act=0, y_act=5.0, alpha=0.5, pair='0x514910771AF9Ca656af840dff83E8264EcF986CA/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2', cid='0x4ffa6a02bd60761adfa89906bf3aa7dbcc862702c1f19385ac1663b0e7222432-0', fee=2000.0, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE 2000', constr='carb', params={'exchange': 'carbon_v1', 'tknx_dec': 18, 'tkny_dec': 18, 'tknx_addr': '0x514910771AF9Ca656af840dff83E8264EcF986CA', 'tkny_addr': '0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', 'blocklud': 19764256, 'y': 5.0, 'yint': 8.529145462309762, 'A': 0.002257868117407469, 'B': 0.06480740698407672, 'pa': 0.004497751124437756, 'pb': 0.004199999999999756, 'minrw': 1e-06}),
ConstantProductCurve(k=14203024.226426456, x=249.22733623585125, x_act=0, y_act=807.9145301701096, alpha=0.5, pair='0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2/0x514910771AF9Ca656af840dff83E8264EcF986CA', cid='0x4ffa6a02bd60761adfa89906bf3aa7dbcc862702c1f19385ac1663b0e7222432-1', fee=2000.0, descr='carbon_v1 0x514910771AF9Ca656af840dff83E8264EcF986CA/0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE 2000', constr='carb', params={'exchange': 'carbon_v1', 'tknx_dec': 18, 'tkny_dec': 18, 'tknx_addr': '0x514910771AF9Ca656af840dff83E8264EcF986CA', 'tkny_addr': '0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', 'blocklud': 19764256, 'y': 807.9145301701096, 'yint': 1957.3030953877023, 'A': 0.519359008452966, 'B': 14.907119849998594, 'pa': 237.97624997025295, 'pb': 222.22222222222211, 'minrw': 1e-06}),
ConstantProductCurve(k=55713831.26523774, x=131.16846526787972, x_act=0, y_act=15920.776548455418, alpha=0.5, pair='0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2/0x8E870D67F660D95d5be530380D0eC0bd388289E1', cid='0xd53681c4f9860ae4353eda47246af69840efbc971f62fa2e30a8426145345f0b-0', fee=2000.0, descr='carbon_v1 0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'tknx_dec': 18, 'tkny_dec': 18, 'tknx_addr': '0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', 'tkny_addr': '0x8E870D67F660D95d5be530380D0eC0bd388289E1', 'blocklud': 19764263, 'y': 15920.776548455418, 'yint': 32646.46821984497, 'A': 4.373757425036729, 'B': 54.77225575051648, 'pa': 3498.2508745627138, 'pb': 2999.9999999999854, 'minrw': 1e-06}),
ConstantProductCurve(k=55692277.270269215, x=424668.0260493125, x_act=0, y_act=5.0, alpha=0.5, pair='0x8E870D67F660D95d5be530380D0eC0bd388289E1/0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2', cid='0xd53681c4f9860ae4353eda47246af69840efbc971f62fa2e30a8426145345f0b-1', fee=2000.0, descr='carbon_v1 0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE/0x8E870D67F660D95d5be530380D0eC0bd388289E1 2000', constr='carb', params={'exchange': 'carbon_v1', 'tknx_dec': 18, 'tkny_dec': 18, 'tknx_addr': '0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE', 'tkny_addr': '0x8E870D67F660D95d5be530380D0eC0bd388289E1', 'blocklud': 19764263, 'y': 5.0, 'yint': 10.072969792409115, 'A': 0.0013497708452092638, 'B': 0.016903085094568837, 'pa': 0.0003331667499582927, 'pb': 0.0002857142857142352, 'minrw': 1e-06})
]
CCRaw = CPCContainer(CurvesRaw)
CCRaw2 = CPCContainer(CurvesRaw2)

# ### >> Enter prices
#
# Provide current prices (`pstart`) here, in the format
#
# PRICES = {
# '0x8E87...': 0.0003087360213944532,
# '0x5149...': 0.004372219704179475,
# '0xEeee...': 1
# }
#
# The price numeraire does not matter as long as they are all in the same numeraire. All tokens must be present. Additional tokens can be added and will be ignored.

PRICES_RAW = {
'0x8E870D67F660D95d5be530380D0eC0bd388289E1': 0.0003087360213944532,
'0x514910771AF9Ca656af840dff83E8264EcF986CA': 0.004372219704179475,
'0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE': 1
'0x514910771AF9Ca656af840dff83E8264EcF986CA': 0.004372219704179475,
'0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2': 1,
#'0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE': 1,
}

# ### >> Enter tokens
#
# Provide token tickers here, in the format
#
# TOKENS = {
# "0x5149...": "LINK",
# "0x8E87...": "USDP",
# "0xEeee...": "ETH",
# }
#
# All tokens must be present. Additional tokens will be ignored. You must also provide the `TARGET_TOKEN` (default: first token of `TOKENS`)
#

# +
TOKENS = {
"0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE": "ETH",
"0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2": "ETH",
#"0xEeeeeEeeeEeEeeEeEeEeeEEEeeeeEeeeeeeeEEeE": "ETH",
"0x514910771AF9Ca656af840dff83E8264EcF986CA": "LINK",
"0x8E870D67F660D95d5be530380D0eC0bd388289E1": "USDP",
}
Expand All @@ -109,18 +86,29 @@

# ### >>> Run optimizer
#
# please make sure that this line runs without errors (other than the error that needs to be addressed of course)
# this run diverges (see the excessive result, 8.69 ETH)

O = MargPOptimizer(CCRaw)
O = MargPOptimizer(CCRaw1)
r = O.optimize(sfc=TARGET_TOKEN_RAW, params=dict(pstart=PRICES_RAW))
print(r.result)
r

# for reference, one the second curve set it converges (see result, 0.047 ETH)

O = MargPOptimizer(CCRaw2)
r = O.optimize(sfc=TARGET_TOKEN_RAW, params=dict(pstart=PRICES_RAW))
print(r.result)
r

# This code runs with a wider Jacobian (`jach=...`). It converges.

O = MargPOptimizer(CCRaw1)
r = O.optimize(sfc=TARGET_TOKEN_RAW, params=dict(pstart=PRICES_RAW, jach=0.001))
print(r.result)
r

# **do not worry about the code below here; this is for the actual testing and will be adapted as need be**

# ### >>> Preprocessing
#
# Please ensure that this code runs without error. Errors here mean that the data provided above is not consistent.

def replace_tokens(dct):
"""replaces the token address with the token name in dct"""
Expand All @@ -132,9 +120,8 @@ def replace_tokens(dct):
return dct


# If this fails this probably means that one of the tokens has not been defined above

CC = CPCContainer.from_dicts([replace_tokens(d) for d in CCRaw.asdicts()])
CC1 = CPCContainer.from_dicts([replace_tokens(d) for d in CCRaw1.asdicts()])
CC2 = CPCContainer.from_dicts([replace_tokens(d) for d in CCRaw2.asdicts()])
PRICES = {TOKENS[addr]:price for addr, price in PRICES_RAW.items()}
TARGET_TOKEN = TOKENS[TARGET_TOKEN_RAW]
PRICES
Expand All @@ -155,6 +142,8 @@ def p(pair=None, *, tknb=None, tknq=None, prices=None):

# ## Curves

CC = CC1

print("Num curves: ", len(CC))
print("Pairs: ", set(c.pairo.primary_n for c in CC))
print("Target token: ", TARGET_TOKEN)
Expand Down Expand Up @@ -208,6 +197,9 @@ def p(pair=None, *, tknb=None, tknq=None, prices=None):

(228.716777/228.602476-1)*10000, (3239.013043/3237.394345-1)*10000

import numpy as np
np.log10(1.01)

# For reference, the CID dataframe `ciddf` (separate because the field is too long; can be joined to `curvedf` via index)

ciddf = pd.DataFrame([dict(cid=c.cid) for c in CC])
Expand All @@ -222,16 +214,34 @@ def p(pair=None, *, tknb=None, tknq=None, prices=None):

# ## MargPOptimizer

O = MargPOptimizer(CC)
r = O.optimize(sfc="ETH", params=dict(
O = MargPOptimizer(CC1)
r = O.optimize(sfc=TARGET_TOKEN, params=dict(
pstart=PRICES,
dump_curves = O.DC_DEFAULT,
#jach = 0.001,
verbose=True,
debug=False,
debug_j=True,
debug_dtkn=False,
debug_dtkn2=False,
debug_dtknd=True,
))
print(r.result)
r

O = MargPOptimizer(CC1)
r = O.optimize(sfc=TARGET_TOKEN, params=dict(
pstart=PRICES,
dump_curves = O.DC_DEFAULT,
jach = 0.001,
verbose=True,
debug=False,
debug_j=True,
debug_dtkn=False,
debug_dtkn2=False,
debug_dtknd=True,
))
print(r.result)
r


Loading

0 comments on commit 768f75a

Please sign in to comment.