Flat betting: a constant stake, longer sessions and the same edge

Flat betting means you pick a unit before the session and you do not change it because the last result was a win or a loss. If the unit is $2, the next bet is $2 after a miss and $2 after a hit. You may still stop. You may still walk. You do not let the notepad, the Fibonacci step or a “due” feeling pick the next number.

That is the whole rule. It looks too plain to be a system, which is why brochures skip it. It is the baseline every progression should be scored against. If a ladder cannot beat a constant unit on the same total dollars, the ladder is entertainment, not arithmetic.

What it is not

• It is not a claim that the next spin is independent — though that claim is true, and gambler’s fallacy is the usual argument against it.
• It is not the Kelly criterion. Kelly changes the fraction of bankroll as the estimated edge and the bankroll change. On a house game the estimated edge is negative, so full Kelly says bet nothing.
• It is not “betting small”. A flat $200 is still flat. Longevity is measured in units, not in the sticker on the chip.

The Games and odds topic is full of named ladders. Flat betting is the control group. PVPspinArena is 18+.

Write the same 2x bet, same p, same dollars of action. Only the *order* of stake sizes changes.

None of these plans changes p or the payout. They allocate the same taxed dollars to different moments. Negative progressions put extra dollars on the table after you are already losing. Positive progressions put extra dollars on the table after you are already winning. Flat betting refuses the argument and pays the edge at a constant size.

Worked order effect

Five-bet tape on a 2x colour, unit $1: L, L, W, L, W.

• Flat: −1 −1 +1 −1 +1 = −$1. Five dollars of action.
• Martingale starting at $1: −1, −2, +4, −1, +2 = +$2 on $10 of action — this particular order paid. Swap the last W for an L and the same plan is already in a hole looking at $2, $4, $8.
• Labouchere 1-2-3: first stake is $4, so this five-bet window is a different experiment entirely.

Order luck is not a theorem. The brochure always shows the order that flatters the ladder. Flat betting makes the brochure boring, which is the point.

Longevity is how many decisions a bankroll can fund before it can no longer place the next bet. Progressions spend that budget on a few large tickets. Flat betting spends it on many small ones.

Units, not dollars

$200 at $10 a bet is 20 units. $200 at $2 a bet is 100 units. Ruin talk is in units. The risk of ruin guide is the formula page; this page is the staking consequence.

On a fair even-money bet, starting with i units and aiming for N, P(ruin) = 1 − i/N. Twenty units aiming to double (N = 40) has a 50% chance of dying first. The same 20 units under a Martingale that doubles from 1 can die in a handful of losses: 1+2+4+8+16 = 31 already exceeds 20, so five misses ruin you. Probability of five opening misses on a fair coin is 1/32 ≈ 3.1% *per start*, and you take many starts.

On Purple, q = 8/15. Five opening misses: (8/15)^5 ≈ 4.3%. The extra slots make the Martingale wreck more common. Flat $1 on the same 20-unit roll is still slowly leaking 6.67% per dollar, but the session usually ends because you chose a time or loss cap, not because five reds ate the stack.

Expected leak versus time at the table

A 100-bet flat session at $1 on Purple has expected loss ≈ $6.67 and a standard deviation around $10 (even-money variance is order 1 per bet, so √100 = 10). Finishing +$10 or −$20 is ordinary. A Martingale session that “wins $1 eight times then dies” can print +$8 then −$20 on the ninth start. Same edge per dollar; different chance you are still in the room after an hour.

If the goal is a long sitting with a known hourly cost, flat is the plan that makes the hourly cost visible. If the goal is a small advertised plus with a hidden tail, that is Martingale marketing.

House edge is a percentage of amount wagered. Flat, ladder or pyramid, the mean is edge × turnover.

On Roulette, Purple and Silver pay 2x with 7/15 hits. Return is 14/15 ≈ 93.33%, edge 6.67%. One hundred flat $2 bets turn over $200 and expect to leak about $13.33. One hundred messy Labouchere dollars expect to leak the same $13.33. The Labouchere path may finish +$6 on a completed line or −$80 on a wreck; the average of those stories still tracks the $13.

Why people still raise

Raising after a loss is usually an attempt to get even in one bet. Raising after a win is usually an attempt to “press a streak”. Both treat the next independent trial as if it had a memory. A constant stake is the behaviour that matches the maths: the next trial is the same price as the last one.

A positive-edge player is a different animal. A card counter or a plus-EV sports bettor *should* vary the stake with the edge and the bankroll. That is Kelly or a fraction of Kelly, and it is not what this page is about. On a posted house game the edge is against you on every posted bet, so varying the stake is varying how fast you pay it.

Stops are not raises

The useful “system” hiding inside flat betting is a pair of written stops:

1. a loss cap in units (for example, 20 units);
2. a time or decision cap (for example, 50 bets or 40 minutes).

Those stops cut exposure. They do not create an edge. They are the same stops a progression player should have written and usually did not.

Pick the unit from the bankroll, not from the last win.

• Session bankroll, not net worth. The money on the table is the money you can lose tonight without touching rent.
• Units of comfort. A common entertainment range is 50–100 units in the session stack, so a $100 sitting implies $1–$2. Twenty units is already a short fuse on a 6.67% colour.
• Match reality. On Coinflip the other player has to accept the size. A “flat” $50 that nobody will fade is a sitting-out rule, not a staking plan.
• Jackpot shares are not even-money units. A Jackpot ticket’s win chance equals your share of the pot. Flat-staking jackpot entries means repeating a chosen ticket size, not pretending each entry is a 2x coin.

The bankroll calculator is the tool page for sizing. This page is the reason the tool asks for a constant unit.

If a $2 unit already feels like it must be recovered tonight, the unit is too large. Flattening the stake does not flatten the mood. Use the limits and help links on the responsible gambling page. Gambling is 18+ and optional.

PVPspinArena runs three player-vs-player games. The same constant-stake idea maps cleanly onto two of them and loosely onto the third.

• Roulette. Purple or Silver is the even-money canvas. Write $X a spin, write a 20-unit stop, and let the 6.67% edge be the visible hourly price. Green is a 14x long shot; a “flat” green unit is a different variance product, not a colour-betting system.
• Coinflip. Two players, 50/50, winner takes the pot minus any fee shown before entry. Flat means repeating a size you can lose several times, not pressing after a miss because “it has to flip”.
• Jackpot. Win chance equals your share of the pot. Repeating a ticket size is honest. Building a cancellation line on ticket sizes is not.

Results come from committed seeds. After a sitting, any settled round can be checked on fairness. Verification answers “was this tape honest?” It does not answer “should I have raised?” The honest tape plus a constant unit is the clean experiment: count how many units left after 30 decisions. That number, not a brochure pyramid, is what a 6.67% colour costs.

Prestige XP on prestige tracks wager volume. A flat unit still writes XP as you play; it just refuses to let volume spike because the last spin missed. If you want a rule because it stops you inventing stakes, keep the rule and skip the folklore ladder.

A last comparison in units: thirty flat $1 Purple bets expect to leak about $2.00. Thirty dollars put through a 1-2-3 Labouchere line can finish as five booked $6 targets or as one wreck. The mean is still the edge on the dollars that actually went down. Flat betting is the plan that makes that sentence visible before you sit.