Position Sizing and Risk Management Basics

Most people who get into algorithmic trading spend nearly all their effort on the entry signal — the rule that says when to buy. It’s the fun part. But the signal is not what determines whether an account survives. Position sizing and risk management do that, and they matter more than the signal, because a good edge sized recklessly still goes to zero, while a mediocre edge sized carefully can quietly compound.

This is educational material, not financial advice. Crypto trading is high-risk, most retail algo traders lose money, and nothing here is a recommendation to trade.

The asymmetry of losses

Start with the single most important fact in risk, because it reshapes everything: losses and gains are not symmetric.

Lose 10% and you need about 11% to get back to even. Lose 50% and you need 100% — you must double what’s left. Lose 90% and you need a 900% gain to recover. The deeper the hole, the more brutally the math turns against you.

The practical consequence is that avoiding large drawdowns matters more than capturing large gains. A strategy that never loses more than a manageable fraction gives compounding room to work. One that occasionally takes a huge loss spends most of its life climbing out of holes. Protecting the downside isn’t caution for its own sake; it’s arithmetic.

Position sizing is the real control

Your position size — how much capital you commit to a single trade — is the lever you actually control. The market decides whether a trade wins. You decide how much it costs you if it loses. Getting sizing right is most of risk management.

A few common approaches, from crude to more principled:

Fixed fractional. Risk a fixed percentage of the account on each trade — say, some small fraction f of equity. Because it’s a percentage of the current balance, you automatically bet less after losses (protecting capital) and more after gains (compounding). Keeping f small is what keeps a losing streak survivable.

Risk-based sizing with a stop. More precise: decide in advance where you’d exit if wrong (your stop), then size the position so that hitting the stop costs only your chosen risk fraction. If you’re willing to risk R dollars and your entry-to-stop distance is d per unit, you buy R / d units. This ties size to how far the trade can go against you before you admit it’s wrong, rather than to a fixed dollar amount.

Volatility targeting. Size inversely to the asset’s volatility so each position contributes a similar amount of risk. A calm asset gets a larger allocation; a wild one gets a smaller one. This stops a single high-volatility coin from dominating the portfolio’s risk just because you allocated the same dollars to it. In crypto, where volatility varies enormously between assets and over time, this idea does a lot of work.

A word on the Kelly criterion, which you’ll encounter: it gives a mathematically “growth-optimal” bet size given your edge and odds. In practice full Kelly is far too aggressive for real trading — it assumes you know your edge precisely (you don’t) and tolerates gut-wrenching drawdowns. Most people who use it at all use a small fraction of it. The lesson isn’t the formula; it’s that there is such a thing as betting too big even when you have an edge.

Think in risk, not dollars

A mindset shift that separates durable traders from the rest: measure everything in risk, not notional dollars.

“I bought $5,000 of a token” tells you almost nothing. “This position risks about 1% of my account if my stop is hit” tells you what you actually need to know. When every position is expressed as a fraction of account risk, you can reason about the portfolio as a whole — total risk on, correlation between positions, worst-case if several move against you at once. Correlation matters especially in crypto: assets tend to crash together, so ten “diversified” alt positions can behave like one big bet when the market turns.

Drawdown is the number that ends strategies

Drawdown is the drop from a peak in equity to the following trough. It’s worth watching more closely than returns, for two reasons. First, it’s what actually forces people to stop — a strategy you’d have ridden through on a spreadsheet is one you abandon at a real 40% loss because you can’t sleep. Second, it’s a signal that a strategy may have stopped working: a live drawdown deeper or longer than anything in your backtest is evidence the market has changed, not just noise.

Deciding in advance the maximum drawdown you’ll tolerate before halting — and honoring it — is itself a risk control. It converts a vague, emotional “should I quit?” into a rule you set while calm.

Simple rules that prevent ruin

Beyond sizing, a few plain constraints do most of the protective work:

Cap risk per trade at a small fraction of the account, so no single loss is fatal.
Cap total risk on across all open positions, so a correlated cluster can’t sink you at once.
Use stops or a defined exit so a losing trade has a floor, and respect them — a stop you override isn’t a stop.
Be extremely careful with leverage. It multiplies both directions, and in volatile crypto it turns an ordinary drawdown into a liquidation. Many blowups are not bad strategies; they’re fine strategies run at reckless leverage.

The takeaway

Edge decides whether you win on average. Risk management decides whether you’re still around for the average to show up. The uncomfortable truth is that survival has to come first: no signal is good enough to justify a bet that can wipe you out, because the math of recovery is merciless and the market is patient. Size small, think in risk rather than dollars, watch drawdown honestly, and treat leverage as dangerous by default — and if a strategy still can’t survive that discipline, the problem was never the sizing.