TellHawk

ProofRunner methodology

How the simulator and the rule library work.

Plain English, including what they get wrong. If you are about to pay for a funded challenge, read the limits before you read the pass rate.

What this covers

ProofRunner has two parts that this page explains. The Challenge Simulator replays a simple trading strategy against a firm's challenge limits to show how often it would have passed and how it failed. The rule library holds the limits themselves: profit target, drawdown, daily loss, minimum days, consistency, contracts, news and session rules, for the firms we have researched.

Both are research tools. They are not advice, they do not predict a result, and they are not affiliated with or endorsed by any firm.

How the simulator works

  1. It runs a strategy on real daily prices. You pick a ticker, a history length (1, 3, 5 or 10 years) and a simple strategy: a moving-average crossover, an RSI rule, or a rule you build from indicators. It uses daily price bars only.
  2. It sizes every trade the same way. Each trade risks the percentage of current equity you choose if it loses one "R". One R is the price move you set as your stop distance or, if you leave it blank, the size of a typical losing trade in the sample.
  3. It replays the challenge from many start dates. A fresh challenge starts every 7 days across the history, using the trades that actually followed. This keeps the real order and clustering of trades. Because the windows overlap heavily, the page also reports how many roughly independent windows there really were.
  4. It resamples the trades. A second method reshuffles the trade sequence in short blocks, so streaks of wins or losses stay together, and keeps the waiting time that came before each trade. It runs up to 2,000 resampled challenges. The random seed is fixed on our side, so the same inputs always give the same answer and nobody can shop for a lucky seed.
  5. It reports passes and breaches. You see the share of runs that passed, the share that broke each rule (daily loss, drawdown, time limit, too few trading days) and how deep the account dipped. Where the two methods disagree by more than 20 points, it says so and tells you not to average them.

The evaluation window defaults to 90 days. That is a horizon you choose, not a rule taken from a firm.

Assumptions and limits

Every result opens with a "Read this first" box. These are the same points, in one place:

It is optimistic about losses

  • Daily loss is worked out from closed trades, grouped by exit date (UTC), not from intraday equity. A firm also counts open-trade losses during the day, so real daily-loss breaches are more likely than shown.
  • Drawdown and profit target use equity after each closed trade, with no floating profit or loss in between.
  • Slippage, swap, commissions beyond the stated backtest costs, news restrictions and weekend gaps are not modelled.

Small samples are flagged

  • Each result carries a sample-confidence rating: unreliable below 10 trades; low below 30 trades or with fewer than 3 independent windows; moderate below 100 trades or 8 windows; fair above that.
  • A pass rate from 20 trades is a hint, not a probability. Resampling cannot create information that is not in the trades.
  • If a sample has no losing trades, the result says it is optimistic and unreliable.
  • Futures firms are not simulated yet. Futures price history licensed for customer use is not available yet, so the simulator will not run a futures firm's rule set. It says so instead of substituting other data. For futures, the guardian, failure autopsy and journal work from your own account feed or a statement import.
  • The simulator applies three limits. Profit target, maximum daily loss and maximum drawdown, plus an optional minimum number of trading days. The drawdown floor is static, set from the starting balance. Other rules in the library (consistency, contract limits, news windows, trailing drawdown) are shown there, but this simulator does not apply them.
  • Data coverage is limited. US stocks and ETFs, major indices, a set of futures price series, BTC-USD and daily FX reference rates. Spot-forex symbols are not loaded. FX runs use one reference-rate close per day, with no intraday range and no dealer spread. Market prices come from a third-party provider and can be delayed or wrong.
  • Leverage is not checked. If a position works out to many times your account (the result flags it above about 10 times), check your firm's leverage limit yourself. The simulator does not model it.

How rule data is researched

  • We read the firm's own public pages. Help centres, rule pages and terms. We record the page address and the date we read it next to each value.
  • Where a firm's page would not load for us, we say so and mark the values unverified. We do not fill them in from memory or from a third-party summary and call them confirmed.
  • We show the firm's own wording on bots, EAs and copy trading where we found it, because rules on automation differ between firms and change. Read the firm's page before connecting anything.
  • No endorsements, no affiliate links. A rule set is a modelling aid. We do not rank firms, say which one to buy, or comment on how reliable any firm is. Firm names are used only to describe what the rules say.
  • Every profile can be edited. Copy any rule set and change any number. A rule set you build yourself is always shown as not verified, with the numbers exactly as you entered them.

Confirmed and unverified

Each field in a rule set carries one of two marks:

  • Confirmed means the value was read from the firm's own page on the date shown, with a link to that page.
  • Unverified means we could not confirm it from the firm's own page. It may come from earlier notes, a typical challenge shape, or a source we could not check. Treat it as a starting point to check, not as the firm's terms. The FX and CFD profiles are in this group today.

Even a confirmed value is a snapshot. Your account is governed by the firm's current rules, not ours, and firms change them, sometimes without notice.

Re-checks and the change log

We re-check firms by hand. There is no automatic monitor, so we do not promise a schedule. Instead the app shows how old each check is:

  • Up to 60 days: shown as normal, with the date.
  • Over 60 days: amber, "may be out of date, verify with the firm".
  • Over 120 days: red, with the same warning.
  • Never checked: shown as never verified against the firm's own pages.

Each firm has a version. When we publish a change, the version goes up and a line is added to that firm's Changes list in the rule library: what changed, field by field, the old and new value, the date and the firm page it came from. A re-check where nothing changed moves the "last verified" date without changing the version. Signed-in members can follow a firm and get an in-app notice when its rules change or when its profile goes stale.

Report an error

If a rule looks wrong or out of date, email hello@tellhawk.com with the firm, the rule set, the field and a link to the firm's page that says otherwise. We check it against the firm's own page. If we change it, the fix shows up in that firm's change log with the date and source.

No guarantees

Past or simulated results do not predict challenge outcomes. A simulation shows what would have happened to one strategy on past prices under simplified rules, and a real challenge has costs, timing and rules this page says we do not capture. ProofRunner is a research tool, not advice, and nothing here promises a pass, a payout or a profit. Always verify the rules with the firm before you pay for a challenge.

See it for yourself.

ProofRunner is open by invitation. The product page shows what it does today and what it does not.

About ProofRunner