Future Game Simulation vs ICM
ICM ignores blinds and position; Future Game Simulation adds them back. How FGS refines ICM, where the two models disagree, and which to trust.
On this page · 6 sections
ICM is the workhorse of tournament math, but it has a famous blind spot: it treats the game as if it freezes the moment the current hand ends. No more blinds, no position, no future hands. Future Game Simulation (FGS) fixes exactly that — it plays a few hands forward before falling back on ICM, so it “sees” the blinds coming and values position. The result is a slightly sharper, slightly less risk-averse read than raw ICM.
What ICM leaves out
ICM converts chips to prize-pool equity using only two inputs: everyone’s stack size and the payout structure. That’s its strength — it’s simple and fast — and its weakness. ICM has no concept of the blinds, no idea who’s on the button, and no memory that a short stack must post big blind next hand. It effectively assumes the tournament is paused forever the instant your current decision resolves. For a full refresher on the underlying math, see how ICM is calculated.
That assumption is fine most of the time. But it breaks down in two predictable places: very short stacks, and spots where position or the blinds are about to bite.
What FGS adds back
Future Game Simulation runs the tournament forward a fixed number of hands — one, two, sometimes more — simulating who posts, who acts, and how stacks shift, then applies ICM at the end of that simulated stretch. By playing hands forward, FGS naturally accounts for:
- The blinds. A short stack that will be forced all-in by the big blind next orbit is worth less than ICM says.
- Position. Acting last is worth real chips over a series of hands; ICM assigns it zero value.
- Stack momentum. A stack about to be crippled by posting is discounted; a stack about to fold to the button cheaply is credited.
The deeper the simulation, the more of the future it captures — but also the slower it runs, which is why solvers cap it at a handful of hands.
Where the two models disagree: a short stack
Here’s the clearest divergence. Four players are left, near the bubble, payouts of $50 / $30 / $20 / $0 (a $100 pool). One player is down to a single big blind.
| Player | Chips | Chip share | ICM equity |
|---|---|---|---|
| Big A | 40,000 | 40.0% | $35.15 |
| Big B | 30,000 | 30.0% | $31.84 |
| Big C | 29,000 | 29.0% | $31.46 |
| Micro | 1,000 | 1.0% | $1.55 |
ICM hands the 1,000-chip micro stack $1.55 of equity — as if those thousand chips just sit there. But that stack must post the big blind next hand and is almost certainly all-in blind or busted within an orbit. FGS plays those hands and sees the stack evaporate, so it prices the micro closer to a fraction of that $1.55. The equity FGS shaves off the doomed short stack gets redistributed to the players who’ll collect its blinds — chiefly the ones in position to attack it. Notice the equities still sum to the full $100 pool under ICM; FGS reshuffles the same $100, it doesn’t create new money.
How FGS changes your decisions
Because FGS discounts doomed short stacks and values future position, its practical effect is to make you modestly less tight than pure ICM. A borderline call or shove that pure ICM says fold is sometimes a thin +EV play under FGS, because FGS recognizes that folding just leaves you bleeding into the blinds rather than frozen in amber. The ICM risk premium — the extra equity you demand before committing — is a touch smaller under FGS than under raw ICM.
The effect is real but usually small. For chip-leader and mid-stack decisions far from the blinds, FGS and ICM agree closely. The gap widens the shorter you get and the closer the blinds loom.
When to lean on which
| Situation | Better model | Why |
|---|---|---|
| Bubble, comfortable stack | ICM is fine | Blind/position effects are tiny relative to pay-jump risk |
| Very short (under ~5 BB) | FGS | ICM badly over-credits a stack that’s about to post out |
| Deep-stacked, far from money | Chip EV | Both models converge on chip EV here |
| Final-table shove-fold | FGS if available | Position and blind order swing close spots |
The takeaway
FGS isn’t a rival to ICM — it’s ICM with the blinds and position switched back on. For 90% of decisions the two agree, and ICM’s simplicity wins. But when you’re short and the blinds are bearing down, FGS is the more honest model, and it will nudge you toward slightly more aggression than pure ICM allows. Learn ICM cold first from the ICM hub, then treat FGS as the fine-tuning layer, and fold both into your broader tournament strategy.
Frequently asked
What is Future Game Simulation in poker?
Future Game Simulation (FGS) is a tournament equity model that simulates a set number of hands into the future — accounting for blinds, position, and stack dynamics — before falling back on ICM. It corrects ICM's biggest blind spot: that ICM treats a stack as if the game freezes the instant the current hand ends.
Is FGS more accurate than ICM?
For most spots, yes, but only slightly. FGS captures effects ICM ignores — a short stack bleeding chips to the blinds, positional advantage, the cost of being next to post — so it tends to make you a touch less risk-averse than pure ICM. The gap is largest for very short stacks and small elsewhere.
Should I study FGS or ICM first?
ICM first, always. FGS is a refinement layered on top of ICM, not a replacement, and most modern solvers report both. If you understand why chips diminish in value under ICM, the FGS adjustments make sense; without that foundation they're just numbers.
Where do ICM and FGS disagree most?
On short stacks and around the blinds. ICM credits a tiny stack with meaningful equity as if it never has to post again; FGS discounts that stack because it will be blinded down. FGS also values position and being far from the blinds, which ICM is completely blind to.