SGM insurance, also sold as one leg down cover, is a promotion shape rather than one bookmaker's offer: a same game multi is refunded, usually as a bonus bet, when exactly one of its legs loses. Everything about its value follows from that one word, exactly, and it is easy to price the offer as though it said "at least one".
So what the cover adds is not read off the price of the multi. It turns on how often a bet lands precisely one leg short, and that number is not the one a leg-by-leg calculator gives, because legs inside a single game move together.
What one leg down cover actually pays
Three outcomes, and the promotion only touches the middle one.
- Every leg wins: the bookmaker pays the multi at the price you took. The cover does nothing.
- Exactly one leg loses: the stake comes back, normally as a bonus bet, normally up to a stated cap.
- Two or more legs lose: nothing comes back.
Versions of it carry three conditions in some combination: a minimum number of legs, a minimum price per leg or on the whole multi, and a cap on the refund. The leg minimum and the cap do most of the work, and they are the two people skim past.
Note: the refund is normally a bonus bet, so it is not worth a dollar in the dollar. An Australian bonus bet usually pays the winnings and not the stake, which means its worth depends on what it can be converted to. The bonus bet converter works out what one would keep at the prices you can get, after exchange commission. Value it as the bonus bet it is, never as cash.
An offer that hands the stake back when one leg misses gets described as having no way to lose. It is not: two legs short pays nothing, and from five legs up, on the illustrative inputs below, that is the most likely of the three.
Exactly one, not two
Here is the arithmetic that decides the whole thing. Take illustrative legs that each win 70% of the time, treat them as independent for a moment, and work out the three bands.
For n legs each winning with probability p, the chance every leg wins is p to the power of n, and the chance exactly one loses is n multiplied by (1 minus p) multiplied by p to the power of (n minus 1).
| Legs | All win | Exactly one loses | Two or more lose |
|---|---|---|---|
| 3 | 34.3% | 44.1% | 21.6% |
| 4 | 24.0% | 41.2% | 34.8% |
| 5 | 16.8% | 36.0% | 47.2% |
| 6 | 11.8% | 30.3% | 58.0% |
| 7 | 8.2% | 24.7% | 67.1% |
Illustrative probabilities, chosen to show the arithmetic rather than to describe a real game.
Two things in that table matter more than the rest of this page. On these inputs the chance of landing exactly one short peaks at three legs and falls from there, so adding legs makes the cover fire less often rather than more. And the right-hand column, the band the cover never touches, climbs from 21.6% to 67.1% across the same five rows.
Why the minimum leg count changes the bet
Minimum leg counts on this shape commonly sit at four or more, past the peak in the table above. That is the reason a 4-leg SGM and a 7-leg SGM with identical cover are different propositions rather than the same one at two prices.
The relationship is exact. Divide the exactly-one chance by the all-win chance and the p terms cancel, leaving n multiplied by (1 minus p) divided by p. At four legs on 70% legs that is 4 times 0.3 divided by 0.7, which is 1.71. At seven legs it is 3.00.
So on the 4-leg bet the refund arrives about 1.7 times for every time the multi wins, and on the 7-leg bet about 3 times. The longer multi leans far harder on the refund while the refund fires less often in absolute terms, 24.7% against 41.2%.
Getting the exactly-one number for one game
The table above assumed independence, which is the cross-game case rather than this one. Legs inside a single game are linked, and same game multi promotions works the exactly-one chance out the exact way, from the price of every two-leg combination in the bookmaker's own builder. Those pair prices are the hard part: on a thin market they will not all price.
A two-state model gets close on two assumed numbers instead. Assume the game runs the way the multi needs half the time, each leg then winning 85%, and does not the other half, each leg then winning 55%, with the legs independent inside each state. None of that is measured; it is a shape chosen to be workable. Each leg still wins 70% overall, because half of 85% plus half of 55% is 70%, so the single legs price exactly as they did above and only the joint numbers move.
| Four legs at 70% | All win | Exactly one loses | Two or more lose |
|---|---|---|---|
| Treated as independent | 24.0% | 41.2% | 34.8% |
| Correlated, two-state model | 30.7% | 33.4% | 35.9% |
The exactly-one band, the only band the cover pays on, thins from 41.2% to 33.4%, and 33.4% is the number the valuation below runs on.
The mistake to avoid is taking one row for the price and the other for the refund. A bookmaker's SGM price is already built on the correlated all-win chance, which is why an SGM pays less than its legs multiplied out: 30.7% prices these four legs at $3.26 before any margin, against the $4.16 that 1 divided by 0.2401 gives. See how same game multi odds are calculated for that side of it. Take the correlated price from the bookmaker and the independent exactly-one chance from a leg-by-leg calculator and you have used the pessimistic model for what you are paid and the optimistic one for what you are refunded, which flatters the offer every time.
An illustrative valuation, and the size of its error
The valuation has the same shape as any money-back promotion, the general form of which is in how to value a betting promotion:
value per $1 = P(all legs win) x price + P(exactly one loses) x refund value - 1
Example: illustrative inputs, none of them a real offer. Four legs, each assumed to win 70% of the time (about $1.43 a leg), an offered multi price of $2.60, and a refund bonus bet valued at 70 cents in the dollar, which is an input you choose rather than a published figure. Under the correlated model: 0.3068 x $2.60 is 79.8 cents, plus 0.3340 x 70 cents is 23.4 cents, giving $1.031 back on the dollar, so plus 3.1 cents per dollar staked. Under the independent model: 0.2401 x $2.60 is 62.4 cents, plus 0.4116 x 70 cents is 28.8 cents, giving 91.2 cents, so minus 8.8 cents.
Read those two numbers next to each other. One says bet and the other says do not, and the gap between them, 11.9 cents per dollar, is larger than either answer. The independence assumption is wrong inside one game, so every number on this page is an approximation, and here the approximation error is bigger than the result it produces. That is an uncomfortable thing to publish and it is the truth about this calculation.
What survives is the structure. The win leg on its own returns 79.8 cents per dollar, a 20.2 cent hole, and the insurance leg has to fill it. On these inputs it fills it at a refund value of about 60.6 cents in the dollar, which is the break-even. Below that the bet is negative however the offer is worded.
Risk: Betting involves risk. Bookmakers can restrict or close accounts and void bets, prices move, and a positive expected value bet can still lose. There is no guarantee of profit. See responsible gambling for limits and support.
Where the cap bites, and the stake it sets
The cap is a dollar maximum on the refund, and it is the quietest of the three conditions. A cap below your stake does not trim the offer a little: it cuts the insurance leg in proportion.
Keep every other input from the example and cap the refund at half the stake. The refund is then worth 35 cents in the dollar rather than 70, the insurance leg falls from 23.4 cents to 11.7 cents, and the bet moves from plus 3.1 cents to minus 8.6 cents per dollar. One condition, and the sign changed.
| Refund worth per $1 | Insurance leg | Value per $1 |
|---|---|---|
| 50 cents | 16.7 cents | minus 3.5 cents |
| 60 cents | 20.0 cents | minus 0.2 cents |
| 70 cents | 23.4 cents | plus 3.1 cents |
| 80 cents | 26.7 cents | plus 6.5 cents |
So read the cap as a maximum stake rather than a maximum refund. Up to it the valuation above holds, and the sign of the answer turns over inside ten cents of refund value, between the 60-cent and 70-cent rows. Above the cap there is no cover on the extra dollars at all.
Risk: Betting involves risk. These figures are illustrative averages over many bets, not what one multi pays, and a refund paid as a bonus bet is worth nothing if that bonus bet loses. A cap, a leg minimum or a term you have not read can leave the bet uncovered. See responsible gambling for limits and support.
The conditions that decide whether it pays at all
Five things that remove the cover, in roughly the order they catch people.
The leg count after a void. A void or postponed leg is usually struck out and the multi re-priced on what is left, so a 4-leg bet on a 4-leg minimum can lose its cover without a single leg losing. Void legs in a same game multi covers how that settlement works.
Cashing out. An early settlement normally ends the bet and the promotion with it, including when it happens while exactly one leg is already down.
Excluded markets. Where the cover names qualifying markets, a leg outside that list still counts as a leg and can still lose the bet, while not counting towards the minimum.
Opt-in and one bet per game. A promotion nobody opted into is not on the bet, and a second qualifying multi on the same game is often not covered.
The bookmaker's judgement. Bookmaker terms govern throughout, and the bookmaker decides whether a bet qualifies. No software changes that, and none prevents an account being restricted or closed.
Tip: check the leg count the bet will settle on, not the one it was built with. That is the condition that most often removes this cover, and it is decided after the card is final rather than when the bet goes on.
Where B337 fits
SGM Saver builds and places same game multis for the saver or stake-back offers already on your own bookmaker accounts, on the games you pick, across seven bookmakers and four sports. A multi goes on only after the bookmaker confirms the offer applies to it; otherwise that game is skipped. It uses only offers a bookmaker has already issued to the account and never creates, unlocks or claims one, so with no offer on the account no multi is placed.
The controls are yours: the games it may bet on, the minimum and maximum combined odds, the most legs one multi can have, and the player markets it may use in each sport. It builds the legs inside those limits, or you edit the legs for a game and it places those, re-priced live as the bet goes on. Bets go on before the game starts, it runs on your own computer, and nothing is placed while that is off. It cannot stop a bookmaker restricting or closing an account.
SGM Saver needs Full Automation, which is set up with the team, with the price, including any charges on top of the plan price, confirmed before you pay. Multis placed through B337 use credits, a per-bet usage charge. For free and confidential support call 1800 858 858 or visit gamblinghelponline.org.au.
Risk: Betting involves risk. Bookmakers can restrict or close accounts and void bets, prices move, and a positive expected value bet can still lose. There is no guarantee of profit. A saver offer does not stop a same game multi from losing, and many bookmakers restrict automated betting in their terms. See responsible gambling for limits and support.
For the wider family of these promotions and how the refund shapes differ, see same game multi promotions, and for the automation itself, same game multi automation.