edgekit.challenge

Prop-firm challenge simulation. A prop challenge is a path problem, not an average-return problem, so the only honest answer is Monte-Carlo over resampled paths. Both entry points take daily_pnl_dollars (P&L after sizing, dollars per day) so the breach limits are absolute dollars.

What's inside. The ChallengeRules dataclass and three presets (FTMO_1STEP, CFT_2PHASE, BRIGHTFUNDED), plus two Monte-Carlo entry points: simulate (the pass rate) and days_to_pass (the distribution of days to clear all phases). Both use a block-bootstrap (not i.i.d.) so losing streaks stay intact and the breach probability is honest.

ChallengeRules (frozen dataclass)#

A prop-firm evaluation's hard constraints, all in account dollars.

ChallengeRules(account: float, phase_targets: list[float], daily_loss: float,
               max_loss: float, min_days: int = 4)

Field	Type	Meaning
`account`	float	Account size in dollars.
`phase_targets`	list[float]	Profit target per phase in dollars ([10_000] = single phase; [8_000, 5_000] = two phases).
`daily_loss`	float	Positive daily-loss limit in dollars.
`max_loss`	float	Positive overall max-loss limit in dollars.
`min_days`	int (default 4)	Minimum active days before a phase may be marked passed.

Presets#

All three are $100k accounts with dollar limits, encoding each firm's published rules.

Preset	phase_targets	daily_loss	max_loss	min_days
`FTMO_1STEP`	[10_000]	5_000	10_000	4
`CFT_2PHASE`	[8_000, 5_000]	5_000	10_000	4
`BRIGHTFUNDED`	[8_000, 5_000]	5_000	10_000	3

from edgekit import challenge
challenge.FTMO_1STEP     # ChallengeRules(account=100_000, phase_targets=[10_000], daily_loss=5_000, max_loss=10_000, min_days=4)
challenge.CFT_2PHASE     # two phases: [8_000, 5_000]
challenge.BRIGHTFUNDED   # two phases, min_days=3

# a custom rule set
rules = challenge.ChallengeRules(account=50_000, phase_targets=[4_000], daily_loss=2_500, max_loss=5_000)

simulate#

Block-bootstrap pass rate for a sized daily-P&L stream through the rules — the fraction of n Monte-Carlo paths that clear every phase without breaching the daily-loss or max-loss limits.

simulate(daily_pnl_dollars, rules: ChallengeRules, n: int = 12000,
         rng=None, block: int = 5, phase_days: int = 400) -> float

Param	Type	Default	Meaning
`daily_pnl_dollars`	pd.Series	—	Sized P&L per day, in dollars (breach limits are absolute).
`rules`	ChallengeRules	—	The evaluation's constraints (or a preset).
`n`	int	12000	Number of Monte-Carlo paths.
`rng`	Generator \| None	None	Seeded generator for reproducibility.
`block`	int	5	Block length for the bootstrap (keeps losing streaks intact).
`phase_days`	int	400	Max days allowed per phase before the path is abandoned.

Returns a float pass rate in [0, 1].

simulate.py

import numpy as np, pandas as pd
from edgekit import challenge

# sized daily P&L (dollars/day) — e.g. from sizing.size_to_dd(...)["sized"]
idx = pd.date_range("2020-01-01", periods=400, freq="D")
daily_pnl = pd.Series(np.random.default_rng(0).normal(700.0, 200.0, 400), index=idx)

rate = challenge.simulate(daily_pnl, challenge.FTMO_1STEP, n=12000)
print(f"pass rate = {rate:.1%}")

days_to_pass#

Distribution of days-to-pass (all phases) over the paths that pass — how long a successful attempt typically takes.

days_to_pass(daily_pnl_dollars, rules: ChallengeRules, n: int = 12000,
             rng=None, block: int = 5, phase_days: int = 400) -> dict

Returns a dict:

median — median days-to-pass (NaN if nothing passes).
p25, p75 — 25th / 75th percentile days.
mean — mean days-to-pass.
pass_rate — fraction of paths that passed.
n_pass — count of passing paths.

d = challenge.days_to_pass(daily_pnl, challenge.CFT_2PHASE)
print(d["median"], d["p25"], d["p75"], d["pass_rate"])

!Feed it the SIZED stream, not raw R

Both functions expect dollars-per-day after sizing, because the breach limits are absolute dollars. Passing a raw R-stream (or an unsized dollar stream) makes the daily-loss and max-loss checks meaningless. Size first with sizing.size_to_dd, then simulate.

edgekit.challenge

ChallengeRules (frozen dataclass)#

Presets#

simulate#

days_to_pass#

See also#