MLB Betting Strategy

[omgkyleflavin]

Saw BTB's post on another thread talking about how people should explain their reasoning until they gain credibility and that posting just your play doesn't help at first. Feel bad just posting plays in the BTB contest and I've been doing well in MLB so I'm here to explain.

So I've probably watched 20 baseball games in the last three years combined. I watch football religiously and I'd like to think I know the most about basketball in the world. Anyways, I started betting baseball during last season's playoffs without really knowing much about any team. I ended up going like 24-8 and made a ton of money. I had the Cardinals at like +500 and never hedged, pathetic that I didn't but it worked out.

ANYWAYS, I need some feedback on my betting strategy. I'd like to think that numbers in baseball pretty much tell the story and can tell you for the most part what a players worth and I try to find value looking at the previous box scores using that attitude.

EXAMPLE: Derek Lowe and Pedro Martinez are facing at 10:05 PM ET tonight. In Lowe's last start he gave up 7 hits, walked 4 batters, had trouble throwing strikes, yet only let up 1 run and won the game. In Pedro's last start he gave up 3 hits, walked 1 batter, threw mostly strikes, yet let up 4 runs and lost the game. So now I'm betting tonight's game thinking Lowe is due for some of those runs he some how avoided last game and that Pedro will keep up his good stuff and it will show on the scoreboard this time. Like BTB said, it's better to ride hot streaks in baseball than go against them.

I hope at least someone is following me.

ANOTHER EXAMPLE: I was at work all day and didn't get to make any plays so this might but a force but it should give someone a better understanding of my thinking.

Arizona/San Diego tonight (Kennedy vs Bass)

Kennedy gave up 9 hits, walked 2 batters, but only let up 3 runs and won the game. I would tell you Kennedy has a couple of runs coming his way. I would also tell you Bass has made two relief appearances this year and is going to get shelled.

I could get into more detail about hitting too but this might not even make sense to anyone. Surprisingly it really has worked for me extremely well thus far. I'm off to a really good start this season (3-0 in the BTB contest) so I'm going to keep doing what I'm doing. I'm also watching a lot more baseball games.

[SkyBlues42]

These pitchers are going against different teams the second time around, right? That makes a difference. What if the Orioles touched up CC Sabathia for 10 hits, but scored just 2 runs. Then, CC's next start is against Minnesota.

On the other hand, Liriano allowed 3 hits and 3 runs against Oakland.

So you're saying the play would probably be on Minnesota, because CC should give up more runs this time, and Liriano should give up less?

[Beyond the Bets]

(04-12-2012 08:14 PM)omgkyleflavin Wrote:  Kennedy gave up 9 hits, walked 2 batters, but only let up 3 runs and won the game. I would tell you Kennedy has a couple of runs coming his way. I would also tell you Bass has made two relief appearances this year and is going to get shelled.

Your post was so confusing that I wish you would just go back to posting picks without an explanation.

Kidding, of course.

But can you elaborate on the quoted part above? I'm not sure I'm understanding what you're saying there.

[omgkyleflavin]

If you're pitching bad, you're pitching bad and most MLB players are good enough to take advantage.

BTB: Be prepared for a 10X more confusing post.

Say for every 30 hits a pitcher gives up they should give up 10 earned runs. In Kennedy's previous start he gives up 14 hits but only 1 run. Now the next 16 hits he gives up are going to lead to 9 runs. He pitched awful and avoided giving up a lot of runs but he'll see those runs sooner or later.

Make any sense? It's not really regressing but bad pitching leads to runs scored. If you pitch bad and don't let up a lot of runs, just more of a chance your bad pitching leads to runs next game.

[Beyond the Bets]

What if Kennedy gives up 14 hits and nine runs? Will the next 16 hits lead to one run?

[omgkyleflavin]

Haha well using that logic yes. That was my best example of a pitcher being "due" for some runs.

[Beyond the Bets]

What I'm suggesting, I guess, is that I wouldn't always expect an immediate regression. If Kennedy allowed 120 hits over 15 games and yielded just seven runs, I would call that incredibly lucky and unlikely to happen again over the next 15 games.

I'd feel much more confident in saying that than saying, "OK, Kennedy got shelled last game and allowed just one run. So, next game, if he gets similarly shelled, he'll give up nine."

[omgkyleflavin]

I guess by using that to bet I am expecting an immediate regression, but an immediate regression 60% of the time is fine with me. Chances are most of that regression is going to come when he's pitching bad, and it's fair to say that him getting shelled the previous game means he's pitching bad. I'm not a numbers guy but I'd say the numbers are in my favor.

I also look at the bottom of lineups to play team total overs. If in the previous game a team's 6,7,8,9 hitters all had at least one hit but the top of the order struggled and the team only scored a couple of runs or so I see major value on the over. Your top of the order guys are better hitters and have a better chance of turning it around fast so take yours chances on an over when the bottom of the order is seeing the ball well.

The logic is definitely a little flawed but it's working. Looking at more numbers would probably be a lot more helpful in determining what should actually be a play and what shouldn't be I honestly think this is a good basis and a good place to start when looking at a game.

---

And there's only value when the stats don't tell the true story. Kennedy giving up 16 hits and 9 runs makes sense, that's about what it "should" be. If Kennedy gives up 16 hits and 2 runs, or 4 hits and 9 runs then I'm seeing value.

[DOP]

(04-12-2012 08:45 PM)Beyond the Bets Wrote:  What I'm suggesting, I guess, is that I wouldn't always expect an immediate regression. If Kennedy allowed 120 hits over 15 games and yielded just seven runs, I would call that incredibly lucky and unlikely to happen again over the next 15 games.

I'd feel much more confident in saying that than saying, "OK, Kennedy got shelled last game and allowed just one run. So, next game, if he gets similarly shelled, he'll give up nine."

BTB makes a solid point here. It sounds like you are expecting an immediate correction for 'luck' but remember that players can go entire seasons with bad or good luck. For example, Matt Cain gave up HRs on only 3.7% of Fly Balls in 2011. That percentage was 7.4% in 2010 and 8.4% in 2009. He didn't become a different pitcher during that time or have any other numbers that varied significantly year-to-year. He was just extremely lucky (and maybe thanks to a decent increase in GB%)

To me, it sounds like you are just betting as "go against this guy" but what if San Diego was -110? Would you still go against Kennedy because he is due to give up some runs? I'm not saying you have to develop actual lines or numbers but you should have some type of way to gauge what bets fall into an acceptable range.

To me, if you start betting 'individually,' you will get crushed because i think it leads to paralysis by analysis. As in, "oh Kennedy is going to lose this game because he got crushed" You should be thinking.. would you still bet this game if you were required to bet the game 100 times in a row in the same spot? If your answer is "no" then you shouldn't bet the game now either.

Baseball is all about generalizations to me because it is a grind. Teams do not get 'up' for games. 1 start for an MLB pitcher is less than 1/2 of an NFL game. Would you come out at HT of game 1 of the NFL and say "this team is due to tank the second half of this game?"

I don't mind your strategy but you need to allow a longer period to gather some data. Maybe Kennedy has a 90% strand rate after April and he is starting to get lined like Roy Halladay in his games after going 4-0.. but the underlying data (which can be supported by the Runs per Hits data you mention) suggest that he has been very lucky and he is due to face teams that love to take walks and put runners on base..

I hope this helps.

[Beyond the Bets]

Kyle - In the example above, did the top of the order struggle against an ace or a weak starter? Also, was the starter a righty or lefty? And, will tomorrow's starter be an ace or a weak starter? All of those things must be taken into account. Same with pitcher-batter matchups.

I think it's dangerous to take too much out of such a small sample. For instance, if the 1-through-4 hitters went a combined 0-for-16, that's only 16 at-bats out of, what, 2,275 total? (assuming each hitter has ~575 for the season).

That's a teeny tiny blip on the overall radar.

[DOP]

(04-12-2012 08:56 PM)omgkyleflavin Wrote:  I guess by using that to bet I am expecting an immediate regression, but an immediate regression 60% of the time is fine with me. Chances are most of that regression is going to come when he's pitching bad, and it's fair to say that him getting shelled the previous game means he's pitching bad. I'm not a numbers guy but I'd say the numbers are in my favor.

I also look at the bottom of lineups to play team total overs. If in the previous game a team's 6,7,8,9 hitters all had at least one hit but the top of the order struggled and the team only scored a couple of runs or so I see major value on the over. Your top of the order guys are better hitters and have a better chance of turning it around fast so take yours chances on an over when the bottom of the order is seeing the ball well.

The logic is definitely a little flawed but it's working. Looking at more numbers would probably be a lot more helpful in determining what should actually be a play and what shouldn't be I honestly think this is a good basis and a good place to start when looking at a game.

---

And there's only value when the stats don't tell the true story. Kennedy giving up 16 hits and 9 runs makes sense, that's about what it "should" be. If Kennedy gives up 16 hits and 2 runs, or 4 hits and 9 runs then I'm seeing value.

There is no way you can say how Kennedy is pitching based on 1 start.

Also, unless you have done MLB previously, there is no way you can be confident that any strategy is actually 'working' yet.

You need data that is statistically significant..
if you flipped a coin 10 times and got 'heads' 8 times, would you claim that the coin must be faulty and weighted incorrectly? No way. However, if you flipped 10,000 and got heads 8,000... you must have reason to suspect something.

[omgkyleflavin]

I've been tracking plays this year and I'm trying to use more numbers and will continue to do so. By using the coin example are you suggesting I stop doing what I'm doing?

Maybe in baseball the best example is the most recent one.

And the strategy is working. I'm using the strategy to profit off betting on baseball games and I'm accomplishing that so it's working. I do see what you're saying though.

[DOP]

I would never want anyone to stop doing something they love. I'm just saying when you get heads for the 8th out of 10th time, don't assume "its working" just yet.. you should need a lot more data to feel comfortable.

Everyone needs to start somewhere, just don't bet crazy or become overconfident because you have won so far.. i think baseball breaks more people than any other sport because everyone thinks they can do it (unlike NBA which most people think is unbeatable) but its the biggest grind of them all and you can never "learn" anything, you must about be "learning" so keep at it.

or maybe I'm just saying all this so you drop out of the BTB Best MLB Capper competition.......

good luck!

[Beyond the Bets]

(04-12-2012 09:22 PM)omgkyleflavin Wrote:  I've been tracking plays this year and I'm trying to use more numbers and will continue to do so. By using the coin example are you suggesting I stop doing what I'm doing?

Maybe in baseball the best example is the most recent one.

And the strategy is working. I'm using the strategy to profit off betting on baseball games and I'm accomplishing that so it's working. I do see what you're saying though.

It's "working" over a sample of how many games?

[omgkyleflavin]

(04-12-2012 09:31 PM)Beyond the Bets Wrote:  

(04-12-2012 09:22 PM)omgkyleflavin Wrote:  I've been tracking plays this year and I'm trying to use more numbers and will continue to do so. By using the coin example are you suggesting I stop doing what I'm doing?

Maybe in baseball the best example is the most recent one.

And the strategy is working. I'm using the strategy to profit off betting on baseball games and I'm accomplishing that so it's working. I do see what you're saying though.

It's "working" over a sample of how many games?

Probably 40 or so (don't have exact numbers from last year's playoffs). Maybe I'm just like Lebron and only come up big in the playoffs but I'll continue to work on my baseball handicapping and continue to post/track plays. This feedback was pretty rough, it all makes sense in my head!

[DOP]

Lets say your goal is to hit 55% over 1,000 plays.. would which be very impressive.

Through 40 plays, you have hit 70%.. an amazing run. Out the remaining 960 games, you still need to hit 54.4% to reach your original goal..

We are not saying your strategy is wrong or will lose... We're just saying that good strategies take time to determine if "its working"

[JSpauny]

I agree with what DOP and BTB have discussed already. It sounds like you're expecting immediate regression. That just doesn't happen. Now sure if a guy throws a perfect game in one start, his next start will probably be worse and if a guy gives up 10 runs his next start will probably be better, but those are outliers. If you're looking for game-by-game regression from guys who are hitting, say 25 points higher than their career best BA, it's just not going to happen. Like DOP said, guys can get lucky for seasons at a time. There isn't just a baseline number that guys have to hit or its pure luck and they have to regress back to it.

Last season, Emilio Bonifacio batted .296 for the entire season which was 30 points higher than his previous career high. He had an insane BABIP of .372. A high BABIP is to be expected with his speed, but that is absurdly high. If you would have bet against him (I know we don't bet against hitters but bear with me), knowing that he couldn't possibly sustain that number, you would have lost continually.

Might I suggest reading into some sabermetric articles on regression. You can find them on fangraphs, baseball-reference, baseball prospectus, really anywhere on the internet. They will help explain how regression really works.

Might I also suggest digging into ERA vs FIP vs xFIP. It basically outlines what you're trying to accomplish but gives you 3 numbers to compare instead of looking at box score information. xFIP is a much better representation of expected future performance than plain box score numbers.

[ChiefZilla]

(04-12-2012 08:56 PM)omgkyleflavin Wrote:  I also look at the bottom of lineups to play team total overs. If in the previous game a team's 6,7,8,9 hitters all had at least one hit but the top of the order struggled and the team only scored a couple of runs or so I see major value on the over. Your top of the order guys are better hitters and have a better chance of turning it around fast so take yours chances on an over when the bottom of the order is seeing the ball well.

Love the discussion and love that this is a place where any idea can be thrown out there and it results in genuine responses.

In reading through this thread, my only problem is that your logic seems a little inconsistent. For example, in your quote above, you mention that the "top of the order guys have a better chance of turning it around fast."

That's arguable, but if that's your conclusion, that's fine.

Where it gets inconsistent is in the next part, where you say "the bottom of the order is seeing the ball well."

So, the bottom of the order is seeing the ball well ... and that will continue. The top of the order isn't seeing the ball well ... but that won't continue.

See how that's inconsistent? Using your logic, the bottom of the order should be due for an immediate regression just like the top of the order is.

Hope this helps. And it'll hard to convince you otherwise when your strategy is winning, but best of luck to you as you keep trying to learn (and win).

[mount187]

(04-12-2012 09:38 PM)omgkyleflavin Wrote:  Maybe I'm just like Lebron and only come up big in the playoffs ...

If your plan is to make plays like LeBron in the playoffs, don't bet on any teams playing in Boston, and make sure you don't make any bets in the 4Q of the season.

[SkyBlues42]

(04-13-2012 03:52 PM)mount187 Wrote:  

(04-12-2012 09:38 PM)omgkyleflavin Wrote:  Maybe I'm just like Lebron and only come up big in the playoffs ...

If your plan is to make plays like LeBron in the playoffs, don't bet on any teams playing in Boston, and make sure you don't make any bets in the 4Q of the season.