EdgeForge is a privately held fund trading its own capital. We research, test and run systematic strategies in US equity-index and crude oil futures. Every strategy is defined by fixed rules, validated on data it was never fitted to, and traded with a hard limit on how much it can lose.
We trade centrally cleared futures listed on CME Group exchanges. They trade nearly 24 hours a day, five days a week, with deep liquidity and transparent pricing. Each has a micro contract one tenth the size of the standard one, which lets us set position size precisely to the risk we intend to take.
The benchmark for US large-cap equities and one of the most heavily traded futures contracts in the world.
| Standard | Micro | |
|---|---|---|
| Symbol | ES | MES |
| Per point | $50 | $5 |
| Per tick (0.25) | $12.50 | $1.25 |
The 100 largest non-financial companies on Nasdaq: technology-led, and more volatile than the S&P 500.
| Standard | Micro | |
|---|---|---|
| Symbol | NQ | MNQ |
| Per point | $20 | $2 |
| Per tick (0.25) | $5.00 | $0.50 |
US small-cap equities, which often move differently from the large-cap indices and so add diversification.
| Standard | Micro | |
|---|---|---|
| Symbol | RTY | M2K |
| Per point | $50 | $5 |
| Per tick (0.10) | $5.00 | $0.50 |
The global benchmark for oil. It is driven by supply, inventories and geopolitics rather than by equity sentiment.
| Standard | Micro | |
|---|---|---|
| Symbol | CL | MCL |
| Per point | $1,000 | $100 |
| Per tick (0.01) | $10.00 | $1.00 |
Trading hours for all four: Sunday 17:00 to Friday 16:00, US Central Time, with a daily break from 16:00 to 17:00. Futures are leveraged instruments; position sizes are set by risk, not by the capital the exchange requires as margin.
Almost every strategy a computer search can find looks good on the history it was found on. The work is in establishing which of them are likely to keep working. A strategy has to pass every stage below before it trades.
A genetic search explores combinations of entry rules, trading hours, exits and risk settings over more than ten years of tick-level history; for crude oil, that history includes the best bid and offer at the moment of every trade. Every simulated trade pays commission and slippage.
Walk-forward testing: the strategy is judged only on later periods it was never fitted to, and only on fills simulated tick by tick. Its result must clearly beat two controls: the same search run on a market with any predictable pattern removed, and the same strategy with its entries placed at random.
The most recent year of data is held back from every stage above. It is opened once per set of strategies, against pass and fail criteria written down beforehand, and every opening is recorded. A strategy that fails is dropped, not adjusted and re-tested.
Strategies that pass are combined and sized on drawdown, then run automatically. Live signals are verified against the research version of the same rules, so the strategy being traded is the strategy that was tested.
Every control below is enforced in code. None of it depends on someone remembering to act.
Every entry is sent together with a protective stop order and a profit target, so the most a trade can lose is fixed before it is taken.
No position is held through the daily settlement break or over the weekend. New entries stop before the close, and anything still open is closed automatically.
Each account is sized from the worst peak-to-trough loss its strategies have shown in testing, with a margin for a worse one, so that a normal bad run stays well inside the account's loss limit.
A combination of strategies is chosen on its joint drawdown, not on the sum of its parts' returns, which favours strategies whose losses fall on different days.
A separate process reconciles every account's positions and orders with what each strategy believes it holds, and sends an alert if they ever disagree, if a data feed stalls, or if a system stops reporting.
Trade records and our own recorded market data are backed up off site every hour, encrypted, and the restore process has been tested.
We judge every strategy and every combination of strategies on these measures. None of them is reliable on its own, so they are always read together.
MDD = max over t of ( peak equity up to t − equity at t )
The largest fall from a high point to a subsequent low. It is the number a trading account's loss limit is set against, so it is the first thing we look at. We also measure time under water: how long it takes to recover to a new high.
Calmar = annualised net return ÷ maximum drawdown
How many times a year the strategy earns back its worst decline. It is the objective we select on, because it puts drawdown in the denominator: a strategy that adds return and drawdown together does not improve it. Its weakness is that it rests on a single event, the worst decline, so it is never read alone.
Sharpe = mean(daily P&L) ÷ stdev(daily P&L) × √252
The consistency of the whole daily return stream, annualised. Unlike Calmar it uses every day, not just the worst stretch, so a strategy that lost money three days in five would show it here even with a shallow worst decline.
Sortino = mean(daily P&L) ÷ √mean( min(daily P&L, 0)² ) × √252
The same as Sharpe, but penalising only losing days. A Sortino well above the Sharpe means most of the variability is on the upside. Winning days count as zero in the downside term, not as missing values; leaving them out quietly changes the result.
Daily results of +30, −10, +30, −10: the mean is 10 and the standard deviation 20, so the Sharpe ratio is 10 ÷ 20 × √252 = 7.94. The downside terms are 0, −10, 0, −10, whose root mean square is 7.07, so the Sortino ratio is 10 ÷ 7.07 × √252 = 22.45. Same stream, very different readings, because all of its variability above the mean is gains.
Changing position size scales profit and drawdown by the same factor, so it changes none of these ratios. The only way to improve them is to find strategies whose losses fall on different days.
EdgeForge is privately held and trades only its own money. It does not take outside investment, does not manage money for anyone else, and does not offer advice, signals or any other service. This site describes how the fund operates; it is not an invitation to invest.