Combining ML and ATS betting to optimize Kelly expected BR growth

**u21c3f6** · 05-14-13, 08:04 PM

There is no catch if you have an edge on ML wagers. I, however, would suggest 1/2 Kelly.

Joe.

**brettd** · 05-14-13, 11:16 PM

Would this be considered a common strategy? (Ie: taking ML faves and dogs at the spread for BR growth reasons).

It's just that I've not read/heard anyone to do this. It's not been mentioned at the think tank either.

**NunyaBidness** · 05-15-13, 01:15 AM

Yes if both sides have value, you should bet them both to maximize EG. This situation should come up very rarely in the big 4 sports. I know nothing about Rugby, but you're saying here that the books have no idea how to convert moneylines to a fair spread. If that's true, I wouldn't think its wise to spread that knowledge.

These spots do come up regularly in MMA/boxing, (Mayweather straight up with Robert Guerrero +22.5 two weeks ago for example), but it's impossible to do the same sort of quantitative analysis on those spots.

I believe these spots also come up infrequently in large public money games.

**HeeeHAWWWW** · 05-15-13, 01:15 AM

It's been mentioned a few times. Just one of the inevitable outcomes of Kelly staking - higher bankroll growth for the same edge at lower odds.

**brettd** · 05-15-13, 02:42 AM

Originally posted by NunyaBidness

Yes if both sides have value, you should bet them both to maximize EG. This situation should come up very rarely in the big 4 sports. I know nothing about Rugby, but you're saying here that the books have no idea how to convert moneylines to a fair spread. If that's true, I wouldn't think its wise to spread that knowledge.

These spots do come up regularly in MMA/boxing, (Mayweather straight up with Robert Guerrero +22.5 two weeks ago for example), but it's impossible to do the same sort of quantitative analysis on those spots.

I believe these spots also come up infrequently in large public money games.

Not saying that at all, I'm just comparing the same edge of x at different prices. Whether it be -110, -300 or otherwise. It could be any sport.

**brettd** · 05-15-13, 02:50 AM

Originally posted by HeeeHAWWWW

It's been mentioned a few times. Just one of the inevitable outcomes of Kelly staking - higher bankroll growth for the same edge at lower odds.

Hmmm... I find it surprising that I don't see this fact plastered everywhere.

The EV/EG difference between a 5% edge at -110 vs 5% edge at even just -200 is double. That's seems fairly amazing to me. What's more interesting is the fact that books shade the dog in most instances to cover liabilities against big squares playing the dog side.

**brettd** · 05-15-13, 05:04 AM

Originally posted by NunyaBidness

I know nothing about Rugby, but you're saying here that the books have no idea how to convert moneylines to a fair spread. If that's true, I wouldn't think its wise to spread that knowledge.

Actually I see where you're coming from now and the 'problem' of betting ML faves becomes more apparent.

An edge of x ATS does not automatically maintain itself at lower odds, it diminishes greatly. If we take a 60% ATS play at -110, we have 14.5% edge. To find 14.5% edge at odds of say -400 would be extremely difficult.

No fee lunch it seems.

**HeeeHAWWWW** · 05-15-13, 07:26 PM

Originally posted by brettd

Hmmm... I find it surprising that I don't see this fact plastered everywhere.

I guess so. For me everybody should be modelling EG as much as EV anyway, but most gamblers haven't got past "pick a winner".

**Cookie Monster** · 05-19-13, 07:08 PM

I posted this on another thread, but it got swamped on a tread turned into an extended debate with a moronic troll. (Scooter dixit)

Moneyline vs spread is a very interesting field, and only people with level head can appreciate the advantages and disadvantages of each one. Sadly, many people have prejudices such as "never bet ML chalk, juice will kill you", "ML faves are sure money", "ML live dogs are golden", etc. As always, it is all relative and a -600 may be a bargain or a steep price, just as -10.5 are.

The spread is apparently neat, probabilities are about 50/50, so it is easy to gauge if you are a winner or not. OTOH, the spread is an arbitrary point, and teams usually try to maximize the chance for a straight up win, often losing a deserved ATS win. We all know the basketball late fouling, the football "prevent" defense, resting first team, and so many backdoor covers. As optional said, "the line is just an extra variable to take into account with your capping". You are in fact predicting the median of possible outcomes.

Moneylines have the advantage of both teams trying to maximize the chance to win, making them less "noisy". However, there is a price to pay increasing the variance on both chalk and dog bets.

The distribution curve of the results of spread bets is almost a normal bell. The moneylines have a skewed bell curve, to the left for dogs, and to the right for chalks. The unskewed bell has an inherently lower variance. The dog bets have a downward tendency, compensated for a higher right tail (a chance for a big win). OTOH, the chalk bets have a growth tendency, with a chance for a disaster.

The dog MLs have typically a higher +EV. If you find a +500 fair-price bet offered at +600, you are getting a +16.7% EV. But a chalk paying -500 on a -600 fair price is only +2.86% EV

However, we know that +EV is not the important number, but +EG. And here the upward trend of chalky MLs compensate for the lack of sheer +EV. In the previous paragraph cases, the -500 chalk has a +0.213% EG, almost the same of the +600 dog +0.222% EG.

There is also a psychological aspect in favor of betting chalky MLs: It makes easier the daily grind, having most days as net winners, even if mixed with a few terrible days. I suppose it is hard to bet a few great +EV ML dogs, just to end on red most of the time. Of course, I am not saying it is good to bet chalk MLs just because the upward trend, you must have the +EG.

Getting out of a ML position is easier than a spread. You may lock-in a win or loss just buying the opposite side. However, the spread movement makes hard to do the same. You may end buying/selling points at bad price, or else rooting for a longshot middle or exposed to it.

Getting back to spread vs MLs, there are cases when the decision is clear. In games with small spreads, the need of most books for having all prices rounded to 5 cents results in misprices of a few cents. The 1 has little value on NBA, yet you see it sometimes valued at 10 cents difference between spread and ML. Given a +1-110 or ML+100, the ML is the easy choice. Conversely, -1-110 is better than ML-120

And you should not fall into the trap "these few scores almost never happens, so playing -200 is not worth it". Lets call it the Sawyer trap. A -200 ML is usually -4.5 or -5. If fave wins by 1-2-3-4 (and maybe 5) 16% of times, the ML is worth it. You have to account for expected total score, team variability, late fouling and defense, etc. Discarding the ML offhand is short sighted.

So, choosing between from ML to spread is exactly the same as buying or selling half-points. If the price is right, go for it. The ML has higher variance, the spread has the backdoor cover issue. I am happy of having two choices, both are useful weapons on the arsenal.

**brettd** · 05-20-13, 02:30 AM

That was a good read. Thanks for that.

**xyz** · 05-20-13, 10:34 AM

Just in case you bet both the ML and ATS for the same game when you see an edge in both, you need to remember the ML and ATS are strongly correlated, so you should change your bet size accordingly.

**brettd** · 05-21-13, 12:12 AM

No I never would. Thanks anyway.

**cyberbabble** · 05-26-13, 07:02 PM

I've seen discussions on the forums that the spread/ML are some times out of line for some of the big games in the major sports. I can't give you a link and have never seen a post where someone did the calculations. The claim is that in the Superbowl for example that just before kickoff the situation is good enough for a bet.
The pros don't bet a lot more just because it is a big game, but the public does. The public money causes the relation between pointspread and ML to be distorted. The public heavily bets the favorite on the pointspread because they want to be on the winner. They heavily bet the dog on the ML because they don't like getting .80 for a $1 bet like they would with the fav on the ML.
So sharps like us would be on the dog with the pointspread and the fav with the ML.