Risk of Ruin: The Odds You Blow a Funded Account

Risk of ruin is the probability that losses consume a defined amount of capital before you recover, and on a funded account that amount is the drawdown buffer (the distance to the firm's loss limit), not the account size. Risking 1% of a $50,000 Topstep account puts $500 against a $2,000 buffer, 4 losses of room rather than 100, and at an illustrative 53% win rate with 1:1 trades even a static limit gives a 62% chance of eventually blowing the account, against about 0.0006% if the whole $50,000 were at risk.

Key takeaway

Risk of ruin on a funded account is set by how many full losses fit inside the drawdown buffer, not by the account's headline size. Against a static $2,000 buffer at an illustrative 53% win rate and 1:1 payoff, risking $500 per trade carries a 62% lifetime ruin probability and $100 per trade (20 losses of room) carries 9%, and a trailing limit such as Topstep's raises both.

What is risk of ruin on a funded account?

Risk of ruin on a funded account is the chance of losing the whole drawdown buffer, a thin slice of the headline size. According to Topstep's Maximum Loss Limit page (as of September 2026), the $50K Trading Combine has a $2,000 Maximum Loss Limit (MLL), the $100K $3,000 and the $150K $4,500. Topstep's consistency page sets the $50K profit target at $3,000.

Account	Buffer (MLL)	Buffer as % of account	1% of account	Share of buffer	Losses of room (N)
$50K	$2,000	4%	$500	25%	4
$100K	$3,000	3%	$1,000	33.3%	3
$150K	$4,500	3%	$1,500	33.3%	3

The account size is the wrong denominator. What can actually be lost before the account closes is the buffer, and against the buffer a "conservative" 1% is a quarter of everything.

The common rule of risking 1-2% per trade is a rule of thumb, and it is stated against account size, the number that matters least once a firm sets a loss limit. Sizing mechanics are in the 1% rule rebuilt for funded accounts.

What is the risk of ruin formula?

For independent 1:1 trades with fixed dollar risk and a known win probability p, the risk of eventual ruin is (q/p)^N when p is above 0.5 and 1 when p is 0.5 or below, where q = 1 - p and N = buffer ÷ risk per trade. The formula is the classic gambler's ruin result, derived in Columbia University lecture notes on the Gambler's Ruin Problem, which state what happens without an edge: "if p ≤ 0.5 (each gamble is not in his favor), then with probability one the gambler will get ruined."

An evaluation also ends at the profit target. With the target M units above the start, the probability of hitting the loss limit first is (r^N - r^N+M) / (1 - r^N+M) with r = q/p, or M/(N+M) at p = 0.5; Topstep's $50K target is 1.5 times its buffer, so M = 1.5N.

The formulas in this section assume independent trades, a constant known p, fixed dollar risk and exact 1:1 outcomes, so real ruin for an estimated p is usually higher. The win rates below are illustrative, with dollar risk shown for a $2,000 buffer.

Win rate p	N = 4 ($500/trade)	N = 5 ($400)	N = 10 ($200)	N = 20 ($100)
50%	1	1	1	1
53%	0.618	0.548	0.301	0.0905
55%	0.448	0.367	0.134	0.0181
60%	0.198	0.132	0.0173	0.000301

Doubling the room squares the ruin probability, which is why cutting risk per trade does more than improving win rate at the margins. At 50% no amount of room helps: ruin is certain eventually.

Doubling N squares the ruin probability: at a 55% win rate, 10 losses of room carry 13.4% ruin and 20 carry 1.8%. Between win rates of 51% and 60%, ruin with 4 losses of room falls only from 0.852 to 0.198, while with 20 it falls from 0.449 to 0.000301. No N rescues a 50% trader.

How much should you risk per trade on a 50K account?

On a Topstep $50K account, risking $100 to $200 per trade (5-10% of the $2,000 buffer) holds an illustrative 53% win-rate, 1:1 trader's chance of failing before the $3,000 target to 8.8-26.4%, while $500 per trade raises it to 45.4%.

Risk per trade	% of $50K	% of buffer	N	Fail before target, p = 53%	Fail before target, p = 55%
$50	0.1%	2.5%	40	0.00818	0.000327
$100	0.2%	5%	20	0.0882	0.0180
$125	0.25%	6.25%	16	0.139	0.0400
$200	0.4%	10%	10	0.264	0.129
$250	0.5%	12.5%	8	0.321	0.186
$500	1%	25%	4	0.454	0.362
$1,000	2%	50%	2	0.527	0.478

The same 53% trader fails the evaluation 8.8% of the time at $100 per trade and 45.4% at $500. Nothing about the strategy changed; only the number of losses the buffer could absorb.

Small risk buys safety with time: at a 53% win rate the expectancy is 0.06R per trade, so expected profit reaches the $3,000 target only after 500 trades at $100 risk and 100 trades at $500 (see how the profit target and drawdown interact).

Losing streaks are what spend N. At a 47% loss rate, a given run of four trades loses all four only 0.47⁴ = 4.9% of the time, yet the probability of at least one run of four or more losses is 40.5% within 20 trades, 75.5% within 50 and 94.4% within 100. A 4-loss buffer will sooner or later meet a 4-loss run, so survival depends on profits having moved the balance away from the limit first.

Is 2R better than 1R for surviving a drawdown?

A 2R payoff (win two units, lose one) survives a drawdown better than 1:1 only if its win rate holds up, because ruin depends on expectancy relative to variance, not on reward-to-risk alone. A 1R system winning 55% (+0.10R per trade) and a 2R system winning 40% (+0.20R, twice the expectancy) carry almost identical ruin with 4 losses of room: 0.448 and 0.458. The ratio 2μ/σ² (twice the expectancy over the variance) explains the tie: 0.202 for the 1R system and 0.185 for the 2R system, whose variance of 2.16R² is more than double the 0.99R² of 1:1 trading.

The 2R rows use the exact closed form ρ^N with ρ = (-p + √(p² + 4pq)) / (2p), break-even at p = 1/3; the 1.5R rows solve the walk exactly in half-R steps (+3, -2), break-even at p = 0.40. All values are infinite-horizon ruin at illustrative win rates.

Payoff	Win rate	Expectancy	N = 4	N = 5	N = 10	N = 20
2R	35%	+0.05R	0.820	0.780	0.609	0.371
2R	40%	+0.20R	0.458	0.377	0.142	0.0203
2R	45%	+0.35R	0.259	0.185	0.0341	0.00116
2R	50%	+0.50R	0.146	0.0902	0.00813	0.0000661
1.5R	42%	+0.05R	0.757	0.708	0.508	0.262
1.5R	45%	+0.125R	0.501	0.424	0.186	0.0357
1.5R	50%	+0.25R	0.253	0.181	0.0348	0.00130
1.5R	55%	+0.375R	0.127	0.0769	0.00640	0.0000452

Equal expectancy exposes the variance cost: at +0.05R per trade with 10 losses of room, 1.5R at a 42% win rate carries 0.508 ruin and 2R at 35% carries 0.609. How target distance shifts win rate is covered in R-multiples for funded traders.

How do commissions and slippage change risk of ruin?

Commissions and slippage raise risk of ruin by lowering your effective win rate, and the damage grows as the stop tightens because costs are fixed per contract while risk per contract shrinks. With gross risk R per contract and round-turn cost c on a 1:1 trade, break-even win rate = (R + c) / (2R), and the cost-adjusted win rate is p' = p - c/(2R).

The example uses the Micro E-mini S&P 500 (MES), which CME Group specifies at $5 times the index with a 0.25-point tick worth $1.25, an illustrative round-turn cost of $2.25 per contract ($1.00 of fees plus one tick of slippage), and a 53% raw win rate held fixed across stop sizes (also illustrative).

Stop	R per contract	Break-even win rate	Effective win rate (53% raw)	Net expectancy per contract	Costs as share of gross edge
4 points	$20	55.625%	47.375%	-$1.05	187.5%
5 points	$25	54.5%	48.5%	-$0.75	150%
10 points	$50	52.25%	50.75%	+$0.75	75%
20 points	$100	51.125%	51.875%	+$3.75	37.5%

A 53% win rate is a losing strategy on 4- and 5-point stops once costs are counted, because costs eat 150% or more of the gross edge. Wider stops with fewer contracts keep the same dollar risk and let the edge survive.

On the funded account, 10 MES on a 10-point stop risk $500 gross: a win nets $477.50 (10 × $47.75) and a loss costs $522.50 (10 × $52.25), cutting expectancy from $30.00 to $7.50 per trade. Computed exactly on a $1.25 grid, the chance of losing the $2,000 buffer before the $3,000 target rises from 0.454 to 0.551, and infinite-horizon ruin from 0.618 to 0.875. The same $500 as 20 MES on 5-point stops fails before target 0.660 of the time (with negative expectancy, eventual ruin is certain), and as 5 MES on 20-point stops, 0.500. Cut contracts, not stop distance, to shrink risk per trade.

What does the Kelly criterion say about position size?

The Kelly criterion, the fraction of capital per bet that maximises long-run growth, sizes a 53% win-rate 1:1 trader at 6% of the capital at stake, and on a funded account the capital at stake is the $2,000 buffer, so full Kelly is $120 per trade. The criterion comes from Kelly (1956); Thorp (2006) gives the form for uneven payoffs, f* = (bp - q)/b, where b is the payoff ratio, and works through p = .53 as an example, finding that capital is driven toward zero once the fraction exceeds about 12%. For a 2R system winning 40%, f* = (2 × 0.40 - 0.60)/2 = 0.10, or $200 of a $2,000 buffer.

The 1%-of-account trade ($500, 25% of the buffer) is 4.17 times full Kelly at a 53% win rate and 2.5 times at 55%. At that fraction the expected log growth per trade, p·ln(1 + f) + q·ln(1 - f), is negative: -0.0169 at 53% and -0.0067 at 55%, because 25% sits above the zero-growth fractions of 11.97% and 19.87%. A 25% bet grows only above a 56.3% win rate; at 60% (full Kelly 20%, or $400) it grows +0.0188 per trade. Kelly proper stakes a fraction of current capital, so these multiples describe the first trade.

Fractional Kelly hedges an overestimated edge. Thorp shows that half Kelly keeps three quarters of the growth rate and, in a continuous-time approximation, cuts the chance of ever losing half the starting capital from 1/2 to 1/8. In fixed dollars at a 55% win rate, full Kelly on the buffer ($200, N = 10) carries 0.134 infinite-horizon ruin and half Kelly ($100, N = 20) carries 0.0181.

Kelly (1956) names the theory's limits: "The essential requirements for the validity of the theory are the possibility of reinvestment of profits and the ability to control or vary the amount of money invested or bet in different categories." A trailing loss limit breaks the first requirement, since profits lift the floor instead of enlarging the buffer until the limit locks.

Does risk of ruin apply to a trailing drawdown?

Risk of ruin applies to a trailing drawdown with worse odds, because each new high lifts the loss limit and resets the room to N units below the peak. Topstep's Maximum Loss Limit page describes its version: "The MLL is a trailing limit. It rises as your end-of-day balance grows, but never moves down." Once the limit reaches the starting balance, it locks permanently.

The table compares the static before-target formula with an exact dynamic program for a limit that trails the end-of-day peak and locks at the starting balance, assuming one trade per day and M = 1.5N.

N (risk per trade)	53%: static / trailing	55%: static / trailing	60%: static / trailing
4 ($500)	0.454 / 0.621	0.362 / 0.546	0.183 / 0.358
8 ($250)	0.321 / 0.530	0.186 / 0.386	0.0387 / 0.130
10 ($200)	0.264 / 0.479	0.129 / 0.311	0.0173 / 0.0712
20 ($100)	0.0882 / 0.253	0.0180 / 0.0793	0.000301 / 0.00230

A trailing limit follows your gains up, so the room behind you shrinks exactly when you are winning. It roughly triples ruin at small size, which makes cutting risk per trade even more valuable on trailing accounts.

Trailing hurts most where the static odds look safest: at N = 20 and a 55% win rate, ruin rises more than fourfold, from 0.0180 to 0.0793, because early wins no longer bank a cushion. With no target and the same trail-then-lock rule, lifetime ruin at a 53% win rate rises from the static 0.618 to 0.806 at N = 4 and from 0.0905 to 0.273 at N = 20. A limit that trails forever without locking, with no target to end the game, makes ruin certain at any win rate below 100%, since N straight losses eventually occur. Firm mechanics are compared in trailing vs static vs end-of-day drawdown, and the climb back after a hit is in drawdown recovery math.

Open losses count

Topstep monitors the MLL in real time and counts unrealized P&L, so an open loss that touches the limit fails the account even if the trade would have recovered. The one-trade-per-day model above understates ruin for positions that sink far below entry before closing.

What if you don't know your real win rate?

If your real edge is zero or negative, ruin is certain given enough trades at any position size; the formulas above only price the wait when the edge is known and positive. Base rates are poor: Barber, Lee, Liu and Odean (2014), studying day traders across the entire Taiwan stock market from 1992 to 2006, found that "Less than 1% of the day trader population is able to predictably and reliably earn positive abnormal returns net of fees." Chague, De-Losso and Giovannetti (2020, SSRN working paper 3423101) found that 97% of individuals who began day trading Brazilian equity futures between 2013 and 2015 and persisted for at least 300 days lost money. Neither study covers US prop-firm traders, so applying them here is an inference.

A measured win rate is also noisy. After 100 trades, a 53% result has a standard error of √(0.53 × 0.47 / 100) = 5.0 percentage points, so the 95% range runs from about 43% to 63%; after 500 trades it still spans 48.6% to 57.4%, which includes the 52.25% break-even of the 10-point MES example. Size from the low end of that range (see how long it takes to become profitable and why funded traders fail).

Does copying trades to more accounts reduce ruin?

Copying one strategy to more funded accounts does not reduce ruin, because every copy shares one trade sequence and fails on the same losing run. With N = 4, M = 6 and a 53% win rate, three copied accounts sized alike all fail together with probability 0.454, the same as one account. Three independent strategies with the same individual odds would all fail together only 0.4543³ = 0.0938 of the time, a 90.6% chance that at least one passes against 54.6% for the copied set.

Thor, this blog's own product, is a trade copier, and what a copier adds is execution: the same entries, sized per account so each keeps the same N when buffers differ ($200 per trade on a $2,000 buffer and $450 on a $4,500 buffer both give N = 10). A copier is not the answer while the edge is unproven, because copying a zero-edge strategy spreads a certain ruin across every account and its fees. Prove the edge first, using how to pass a futures prop firm evaluation, then scale.

Go deeper

Frequently asked questions

What is a good risk of ruin percentage?

No regulator, exchange or study sets one; the right level is the failure rate you can afford to pay for repeatedly. At 10%, expect to lose roughly one account in ten attempts with the same edge and sizing, and budget evaluation fees accordingly.

How many losing trades in a row can a Topstep 50K account take?

Three full $500 losses; the fourth touches the $2,000 limit, and touching it fails the account. At $200 per trade the tenth loss ends it, and until the limit locks the count runs from the latest end-of-day high, not the starting balance.

Can you pass a prop firm evaluation with a 50% win rate?

Not reliably at 1:1. With zero edge, the chance of hitting the limit before the target is 60% at every position size on Topstep's $50K ratio of a $2,000 buffer to a $3,000 target, before costs. A 50% win rate needs payoffs above 1:1 after costs to create an edge.

Does hitting the daily loss limit count as ruin?

No, on Topstep only the Maximum Loss Limit ends the account. The $50K Daily Loss Limit of $1,000 is optional; a breach flattens positions and pauses trading until the next session, and the account stays eligible for funding, per Topstep's Daily Loss Limit help page as of September 2026.

What is the difference between risk of ruin and maximum drawdown?

Maximum drawdown measures the largest peak-to-trough fall that already happened; risk of ruin is the forward probability that a fall of a set size happens before recovery. A backtest's maximum drawdown is one sample from that distribution, so a backtest drawdown smaller than the buffer says little about the odds of a larger one.

Are online risk of ruin calculators accurate?

They are accurate for their assumptions and optimistic for real trading. Enter the buffer, not the account, as capital and a cost-adjusted win rate; most calculators still ignore trailing limits, variable trade outcomes and streaky market regimes.

Does trading micros instead of minis lower risk of ruin?

Only by lowering dollar risk per trade. Micros let you set N precisely (one MES point is $5), but commissions are charged per contract, so ten micros usually cost more to trade than one mini of the same size, which lowers the effective win rate.

Does passing the evaluation lower your risk of ruin?

No, the funded stage restarts the count: as of September 2026, Topstep's 50K Express Funded Account begins with its own $2,000 trailing buffer that locks only after $2,000 of profit. The probabilities multiply, so an illustrative 55% chance of passing followed by a 60% chance of reaching the lock leaves 33% overall.

What is gambler's ruin?

Gambler's ruin is the probability problem of a player betting fixed amounts until either reaching a goal or losing the stake, and it is the model behind trading risk-of-ruin formulas. Its central result is that a player without an edge is eventually ruined with certainty when the other side has unlimited capital, which is the market's position relative to one account.