View New Posts
1. ## MLB Series Picks Using Kelly Criteria

I notice quite a few newcomers read this forum. So for the fun of it I thought I'd reintroduce some concepts and apply them to the upcoming MLB series. First is computing the expected returns for a given wager. For this you need 2 pieces of data, the odds and the probability of winning the wager. Looking at the Cleveland - New York series, I found a current line of Cleveland -185 and New York +160. These need to be converted to European style odds for the calculations. Following is the Excel statement that will do that.

=IF(C2>0,(C2+100)/100,(C2-100)/C2)

In this case the odds are Cleveland 1.541 and New York 2.600.

Now for the hard part, assigning the probability of each team winning the series. That's where your handicapping comes into play. I'll cheat and use Nate Silver's fivethirtyeight.com web site to get these. They have Cleveland 67% and New York 33%. Multiplying the odds times the probability gives us the expected return of each wager per dollar bet; Cleveland \$1.032, New York \$0.858.

OK, so the wager is on Cleveland. But how much? For this we'll use the Kelly Criteria. This formula tells you what percentage of your bank roll to wager given the above data. The Excel formula is;

=+(D2*B2-1)/(D2-1)

This suggests a wager of 6.0% on Cleveland and -8.9% on New York. Following is the series table as it exists at the moment.

 538 Prob ML Eur Exp Ret Kelly Indians 67% -185 1.541 \$1.032 6.0% Yankees 33% 160 2.600 \$0.858 -8.9% Red Sox 40% 155 2.550 \$1.020 1.3% Astros 60% -175 1.571 \$0.943 -10.0% Dodgers 64% Diamondbacks 36% Nationals 52% -130 1.769 \$0.920 -10.4% Cubs 48% 110 2.100 \$1.008 0.7%

2.  538 Prob ML Eur Exp Ret Kelly Dodgers 64% -265 1.377 \$0.882 -31.4% Diamondbacks 36% 230 3.300 \$1.188 8.2% Nationals 52% -140 1.714 \$0.891 -15.2% Cubs 48% 120 2.200 \$1.056 4.7%

3. one of the better written basic threads in months.

well done

\$20
Angelman
donation 02/15/2018

4. Good thread. Here is where bettors get into trouble using kelly but it's not a problem with the kelly formula it's the 538 probabilities. Somebody could easily go broke on a busy Saturday using the above probabilities and over estimating their edge.

5. Originally Posted by turbobets
Good thread. Here is where bettors get into trouble using kelly but it's not a problem with the kelly formula it's the 538 probabilities. Somebody could easily go broke on a busy Saturday using the above probabilities and over estimating their edge.
Exactly, well said.

Bsims have you tracked 538's history? How accurate are the projections?

6. Originally Posted by turbobets
Good thread. Here is where bettors get into trouble using kelly but it's not a problem with the kelly formula it's the 538 probabilities. Somebody could easily go broke on a busy Saturday using the above probabilities and over estimating their edge.

7. Hey Bsims, what's the translation of the Excel formula for the Kelly Criteria above?

Is it (Bankroll*ExpReturn - 1) / (Bankroll - 1) ?

8. Originally Posted by StackinGreen
Hey Bsims, what's the translation of the Excel formula for the Kelly Criteria above?

Is it (Bankroll*ExpReturn - 1) / (Bankroll - 1) ?
Kelly = (Odds * Pct - 1) / (Odds - 1)

For Indians it was (1.541 * .67 - 1)/ (1.541 - 1) = 6.0%

9. Originally Posted by oilcountry99
Bsims have you tracked 538's history? How accurate are the projections?
I have not tracked 538. I was just using them as an example because they did have probability for the series.

Several years ago I tracked dozens of touts for an entire college basketball season and concluded that they weren't worth the effort.

Also, I've used Kelly for entire MLB seasons and found that I would be better off with flat bets. Let me look back at these and I'll explain in a separate thread.

10. Originally Posted by Bsims
Kelly = (Odds * Pct - 1) / (Odds - 1)

For Indians it was (1.541 * .67 - 1)/ (1.541 - 1) = 6.0%
Meaning you wager 6% of your bankroll?

11. Originally Posted by Bsims
I have not tracked 538. I was just using them as an example because they did have probability for the series.

Several years ago I tracked dozens of touts for an entire college basketball season and concluded that they weren't worth the effort.

Also, I've used Kelly for entire MLB seasons and found that I would be better off with flat bets. Let me look back at these and I'll explain in a separate thread.
It has been shown on other threads that if you are a better handicapper, you make more money by parlaying. You need to approach a 5% edge though on the market. It's why I've crushed bookies on baseball doing round robins for the last 2 years.

12. Originally Posted by StackinGreen
Meaning you wager 6% of your bankroll?
Yes.

13. My initial posts here used fivethirtyeight.com's probabilities, the Kelly criteria, and odds at the time to suggest the team and size of wagers for the MLB Division Playoffs. Unfortunately, 3 of the 4 suggestions turned out to be losers, with only the Cubs winning. Here in the same format are the suggestions for the League Championships.

Someone asked me about my view of fivethirtyeights's probabilities and I said I was using them only as an example. But here I would agree with the picks of Cubs and Yankees. But that is based on a presentation made at SABR several year's ago. A fellow had done a study and concluded that the post season series were too short for the best team to rise to the top. Hence, I almost always take the dogs.

 National 538 Prob ML Eur Exp Ret Kelly Dodgers 60% -190 1.526 \$0.916 -16.0% Cubs 40% 165 2.650 \$1.060 3.6% American 538 Prob ML Eur Exp Ret Kelly Asros 57% -170 1.588 \$0.905 -16.1% Yankees 43% 155 2.550 \$1.097 6.2%

14. In my view, it should be Dodgers nearing 70-30, the Cubs are outmanned and are far more spent. My line would be -220, or 68.75%

I know it's baseball, but the Dodgers have a better lineup, are actually hitting right now, and you can argue every single one of their starters is superior, with a more rested bullpen.

The Cubs won a coin flip series against a choke team that did that in basically every way during that same series, hitting, pitching, and managing.

Even though I think Houston will win, they aren't that much better than the Yankees (right now) than the Dodgers are Cubs. Surprisingly, the Dodgers I find to be the value line.