The honest answer: less than you think to start, more than you think to make it meaningful. Let me break down the actual math instead of giving you a single number.
The floor: why $2,000 is a practical minimum
Position sizing needs room to work. At 0.5% risk per trade on a $2,000 account, each trade risks $10. With micro lots (0.01) available at most MT5 brokers, that is workable for tighter-stop systems like mean reversion. Below $2,000, minimum lot sizes start forcing you to risk more per trade than the system intends, which quietly changes the risk profile you paid for.
The tiers
| Account | Risk/trade at 0.5% | Realistic use |
|---|---|---|
| $2,000 to $5,000 | $10 to $25 | Mean reversion system, conservative settings, treat it as a learning account |
| $5,000 to $25,000 | $25 to $125 | Full System viable, returns become meaningful in dollar terms |
| $25,000 to $100,000 | $125 to $500 | The zone where systematic trading starts competing with income |
| $100,000+ | $500+ | Our own live account runs here. Compounding does the heavy lifting |
The mistake almost everyone makes
Undercapitalized traders compensate with risk. They take a $1,000 account and run 3% or 5% per trade to make the dollars matter, then one normal losing streak (which every system has) cuts the account in half. The system did not fail. The sizing did. A 10-trade losing streak at 0.5% risk costs about 5% of the account. At 5% risk it costs about 40%.
The prop firm alternative
If your capital is genuinely limited, prop firm challenges flip the equation. For a few hundred dollars in fees you trade a $50,000 or $100,000 funded account and keep 80% or more of profits. The catch is strict drawdown rules, which is why we built a challenge simulator that tests our systems against FTMO-style rules across 11 years of data. At 0.5% risk per strategy the historical pass rate is 100%. That is a capital-efficient path that did not exist a decade ago.
The real answer
Start with what you can lose without changing your life. Run conservative risk. Let the compounding and your monthly savings contributions do the work. An account that survives is worth infinitely more than an account that sprints.
The dollar math at each tier, worked out
Abstract percentages hide what the experience actually feels like, so here is a year at 0.5% risk per strategy on the combined system, at three account sizes, using the backtest's average year as the baseline. On $2,500: a good month adds roughly $85, a normal losing month costs $50, and the year ends around $3,300. Real but small; the value at this tier is proving the process. On $25,000: good months add $850, the max historical drawdown at this risk level is about $1,750 of temporary decline, and the year ends near $33,000. This is where the compounding starts being visible. On $100,000: the numbers scale to $3,400 monthly averages and the year ends around $133,000, at which point the system is competing directly with salaried income while consuming 20 minutes of weekly attention.
The percentages are identical in all three rows. Only the meaning changes, and the meaning is what determines whether you interfere with the system. Which leads to the real point of capital sizing.
The psychological capital requirement nobody prices in
The binding constraint is rarely the broker minimum. It is the size at which drawdowns stop being tolerable to you specifically. A -7% drawdown on $2,500 is $175 and provokes a shrug. The same -7% on $250,000 is $17,500, more than many monthly salaries, evaporating over a few weeks while the correct action is to do nothing. Traders who scale capital faster than they scale drawdown tolerance end up intervening at the bottom, converting the system's temporary decline into their permanent loss. The practical rule: fund the account at whatever size lets you watch the backtest's maximum drawdown happen in dollars without touching anything. Grow from there by compounding rather than deposits, because gains you watched accumulate are psychologically easier to risk than fresh capital. Capital adequacy, it turns out, is mostly emotional adequacy with a dollar sign.