The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →A 96% return-to-player (RTP) figure does not protect your bankroll. It describes the average return a game is designed to produce over a very large number of plays, measured against total stakes. It says nothing about whether your balance will reach zero during a session, how quickly it will get there, or how likely it is to survive a given number of bets. Those outcomes depend on the game’s full payout distribution, how large each wager is relative to your bankroll, and when you decide to stop. This article explains what 96% RTP does and does not measure, shows the classical gambler’s ruin model and where it stops applying to real games, and sets out how to build a simulation whose results you can actually interpret.
What a 96% RTP measures
RTP is a theoretical long-run average. A game with 96% RTP is designed so that, across a sufficiently large number of plays under its modeled behavior, returned winnings equal 96% of the stakes wagered. The UK Gambling Commission’s consumer guidance makes this point directly: the RTP percentage is an average achieved over a significant number of game plays, not something that happens each time a machine is played.
As an Amazon Associate I earn from qualifying purchases.
The Commission illustrates the gap with an 85% RTP example. A player should not expect to win an average of 85 pence for every £1 staked during a playing session. The same logic applies at 96%. Four pence in every pound is the long-run expectation, and a single session can land well above or well below it.
Two readings of “96%” are common and both are wrong. The first is “I get 96% of my money back.” The second is “I have a 4% chance of losing.” RTP is neither a return guarantee for your money nor a probability of any particular outcome. It is an average over outcomes that vary widely.
#1 Best Overall
From RTP to expected loss, and why turnover matters
At 96% theoretical RTP, the expected loss is 4% of total stakes. Total stakes, the Commission’s term is turnover, include winnings that you re-stake during play. The Commission defines turnover as total stakes including reinvested winnings, and gross gambling yield as turnover minus wins. That means the amount wagered can be far larger than the amount you deposited, and the 4% applies to the larger figure.
A hypothetical example makes the distinction concrete. Suppose a player deposits £100 and, through repeated re-staking of winnings, places £1,000 in total bets on a 96% RTP game. The expected loss on those bets is 4% of £1,000, or £40. That is an expectation on turnover. It is not a forecast that the player will be left with £60 from their £100, because the path from £100 to the final balance depends on the sequence of results, which the average does not describe.
Why the bankroll sets a boundary: gambler’s ruin
A finite bankroll creates a hard floor. Once the balance reaches zero, the modeled wagers stop. This is the core of the gambler’s ruin problem: an absorbing boundary at zero that ends the process no matter how the game would have continued.
Recommended Free Tools
The classical model
The textbook version is a simple random walk on the integers from 0 to N. Each step moves the bankroll up by one unit with probability p or down by one unit with probability q = 1 − p. Start with i units, where 0 < i < N. The walk is absorbed when it reaches 0 (ruin) or N (target).
- If p = q = 1/2 (a fair walk), the probability of reaching N before 0 is i/N.
- If p ≠ q, the probability of reaching N before 0 is ((q/p)i − 1) / ((q/p)N − 1).
- The probability of ruin before reaching N is one minus the corresponding value.
These results assume independent steps, fixed one-unit changes, and fixed boundaries. Northwestern University’s Markov chains text treats gambler’s ruin exactly this way, as an absorbing random walk. It is a teaching model for how boundaries and drift interact. It is not a slot-machine formula.
Illustrative numbers
The table below uses hypothetical walks to show how sensitive the answer is to drift. The biased case uses p = 0.49, which is not a 96% RTP game and is chosen only to show the effect of a small bias.
Rank #3
| Scenario | Start (i) | Target (N) | Step probability | Chance of reaching target | Chance of ruin |
|---|---|---|---|---|---|
| Fair walk | 20 units | 40 units | p = 0.50 | 50.0% | 50.0% |
| Slightly unfavorable walk | 20 units | 40 units | p = 0.49 | about 31% | about 69% |
Two things stand out. Even a small unfavorable drift moves the odds of reaching the target from one-half to about one-third in this setup. And the outcome depends on the starting position relative to the target, not only on the drift. Doubling the target while keeping the start fixed changes the answer again, which is why a bankroll expressed in absolute currency means little without the wager size and goal attached.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why RTP cannot produce a ruin probability
RTP specifies a mean. It does not specify how outcomes are spread around that mean. Volatility describes that spread. The Commission describes high-volatility games as having larger tolerances and potentially very large but rare prizes, while low-volatility games tend toward smaller and more frequent wins. Two games with identical RTP can therefore produce very different bankroll paths over the same number of spins.
The table below lists what each input contributes to a ruin estimate.
| Input | What it contributes | What it cannot tell you alone |
|---|---|---|
| Theoretical RTP | Long-run expected return per unit staked | Session outcomes or the chance of reaching zero |
| Payout distribution and volatility | How far actual sessions can depart from the mean, and how often | The expected return on its own |
| Stake per play relative to bankroll | How many losing outcomes the balance can absorb | Whether a long run will be profitable |
| Stopping rule or session length | Whether the path ends at ruin, a target, or a fixed number of plays | The game’s underlying odds |
The Commission’s standards for remote games require operators to provide information about how a game works and about its house edge, RTP, or likelihood of winning. That is one part of describing a game. Volatility and the payout table are the parts needed to model bankroll survival.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to simulate a game’s bankroll paths
- Get the payout distribution. Obtain the game’s outcome probabilities and the amount paid for each outcome. RTP alone is not enough. If the operator publishes only RTP and a volatility label, your model has to be stated as an approximation.
- Fix the stake per play. Use a constant bet in currency units, or define a rule that changes stakes based on the current balance. Each rule produces a different model.
- Set the starting bankroll and the stopping conditions. Define the zero boundary, an optional target balance, and a maximum number of plays.
- Simulate many independent paths. For each path, draw an outcome from the payout distribution, apply the stake and payout to the balance, and stop when the path reaches zero, the target, or the maximum play count.
- Report the endpoints. Record the fraction of paths that ended in ruin, reached the target, or hit the play limit. Report the distribution of final balances too, not only the averages.
- Validate on a known case. Before using the game model, run the simple unit-step walk and compare the simulated frequencies with the formulas above. Agreement within sampling error confirms the code, not the game model.
A minimal simulation of the unit-step walk looks like this. To model a real game, replace the coin flip with a draw from the game’s payout table and multiply the stake by the payout.
Free tools Windows power users keep installed
One-click scans. No signup required.
import random
def simulate(bankroll, target, win_prob, paths=10000, max_rounds=1000):
ruined = reached = 0
for _ in range(paths):
balance = bankroll
for _ in range(max_rounds):
balance += 1 if random.random() < win_prob else -1
if balance <= 0:
ruined += 1
break
if balance >= target:
reached += 1
break
return ruined / paths, reached / paths
ruin, hit_target = simulate(bankroll=20, target=40, win_prob=0.49)
print(f"ruin={ruin:.3f} target={hit_target:.3f}")With 10,000 paths the estimates typically land within a percentage point or two of the formula for this setup, but that tolerance is a property of the sample size, not a guarantee for any game.
Reading simulation output
A simulation answers a conditional question: under these payouts, this stake, this bankroll, and this stopping rule, how often does each endpoint occur? Change any assumption and the answer changes. A model of a high-volatility game with a 500-play limit can show a very different picture from a low-volatility game with a 5,000-play limit, even at the same RTP. For that reason, present results as model estimates for the stated assumptions, not as a prediction for any individual player.
The UK Gambling Commission’s guidance on live RTP monitoring shows how measured results are interpreted. In its example, a game with a 91.68% designed RTP and £1,200,000 turnover that returned £1,085,000 in wins has an actual RTP of 90.42%. The Commission explains that whether such a gap is acceptable depends on volatility and sample size. That example describes aggregate measured performance across a large turnover, not the experience of a single player with a single bankroll.
What regulatory testing does and does not protect
The Commission says games offered online in Great Britain must be tested before release, and that operators must monitor live performance to check fairness and designed RTP. These controls exist to confirm that a game performs as designed. They do not promise that a player’s session will match the theoretical return, and they do not protect a finite bankroll from the zero boundary. A game can be fully compliant and still ruin a player who wagers their whole balance on a high-volatility title.
The clearest protection against gambler’s ruin is structural rather than statistical: set a bankroll you can afford to lose, keep stakes small relative to it, and decide the stopping point before you start.
Quick Recap
“
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
