Why Baseball Handicapper's Hot Streaks are Overrated

There probably is not a day that goes by on internet handicapping forums in which you see bright minds and recreational bettors alike claiming being on “hot streaks” and thinking that they hold substance. Yes, handicappers are humans, much like the players we try to handicap, thus making us subject to variables that affect the quality of our handicapping in the short term. However, that being said, digging deeper into the mathematical aspects of what’s truly behind a hot streak, and one will realize these short term returns that are above our historical means are rather a product of randomness, noise, and good fortunes.

In this example, I would like to use a handicapper that possesses “sharp” like tendencies, as “hot streaks” hold even less substance for the recreational bettors that do not possess an accurate quantitative valuation model to go by. Sharp bettors realize that in reality, they are not betting on teams, rather prices. Therefore, they derive (via power ratings) a valuation model that quantifies their worth of a particular team, and finds disparities between their price and the market price. A positve EV handicapper’s model will find that the results from disparities between “intrinsic” and market prices will lie closer to the perceived value. Therefore, each bet they make will have an “expected positive ROI”, and the more accurate their valuation model is, the closer the actual ROI is to the “expected” ROI in the long run.

For example (simple example). Suppose that there is a quality valuation model that finds 500 games in which derives an intrinsic value of -120 when the market price is -100. If the valuation model generated substance, the results would have produced 273 winners and 227 losers, converting an ROI of 9.2%. Supposing that the handicapper and his (her) model holds long term substance and positive EV, we can diagnose what is truly behind a short term hot streak. Suppose this handicapper goes on a two week hot streak in which he goes on a 30 and 10 run. Is this handicapper hot? Being that his short term ROI of 75% is much larger than his expected 9.2%, he must be “hot” and seeing the sport much easier, right? He must be a good handicapper to follow, at least until his hot streak subsides. Absolutely not. If his valuation model is still generating expected ROI’s of 9%, his excess return is merely noise, or else, it simply defies his models accuracy. The longer his hot streak lasts and the longer his excess ROI is above his expected ROI, the more concerns he should have to the accuracy of his model.

Yes, handicappers are human and the quality of our handicapping changes in the short term. What does this mean exactly? All it means is that if the quality of our handicapping improves in the short term, then the accuracy of the intrinsic value of our lines will improve- not the expected short term returns, as that is a product of other variables as well(the short term market pricing efficiency by linesmakers and simple randomness). So even if the human variables behind our handicapping has a short term structural change effect on the quality of handicapping, the results will more than likely produce a decrease in the standard deviations of our returns due to an increase in model accuracy, not necessarily improved short term results. Simply put, the only reason for an increase in short term returns holding substance for a handicapper is a product of a decrease in market effeciency. What does this mean? This means that the only short term increase in a handicappers ROI that holds substance is a product of an increase of expected ROI due to an increase in the disparity between the intrinsic value and market price.

So yes, hot streaks are nice to have once in a while, as it does improve the size of our bankroll. But the longer they last, the more concerning they should be for an astute handicapper, as a short term ROI of 20% derived from a valuation model that produced an expected ROI of 10% makes the short term accuracy of the model worth as much as if it produced a 0% ROI. So next time you see a handicapper or a tout claiming his hot streak should be followed, please think twice.

is it true you went tout? I heard rumors and wasnt sure if it was true?

Very well written BG. Great read. Thanks for continuing to contribute here.

hahaha, like ur question as the first responce to his "long-awaited" thread

Dsethi said:

This is where I got turned off. Teams are the ones that win and lose money for me, not the prices. Is it important to get a good price? Yes, which is why relying solely on one out isn't always the greatest idea (especially when you're using the Greek for baseball!). But, and even though I don't consider myself "sharp" or even a good capper, I don't agree that sharps bet on prices alone.

And if you perceive a line to be significantly off, isn't that in itself based on the teams that are playing and your perceptions of them and the matchups involved?

Click to expand...

A handicapper should approach a game with the frame of mind that he is willing to be on either team in a given match. Each team has a value (which is predicated and quantified based on all the underlying fundamentals that effect an outcome of a game), and each team has a price (which is the going rate of the best price available for a particular handicappers outs). Bets should solely be based on disparities between the two, or else is defies mathematical reasoning.
<?xml:namespace prefix = o ns = "urn:schemas-microsoft-comfficeffice" /><o> </o>
For example, before I handicapped the Blue Jays/ White Sox game for tomorrow, I was unbiased towards whom I would potentially bet on. I handicapped the game based on all the fundamental factors I perceive fit and derived an intrinsic value for both teams (Jays -174/white sox +174). It was not until the market price came out as to when I formed my bias on Jays, as I perceived them undervalued. So essentially, my bets are predicated on prices, which is a derivative of the team. This takes out all the emotion and psychological barriers in forming biases on teams prior to quantifying their worth.
<o> </o>
I enjoy your posts, as you have a firm grasp on the sport. I think you understood the concept, and it is merely a difference based on semantics rather than approach.

How do you handicap fundamental factors, lines and such?

thecakeoff said:

How do you handicap fundamental factors, lines and such?

Click to expand...

I touched up on the topic in a prior thread. This is what was said.





I thought it might be prudent to share my fundamental approach to handicapping baseball before the season starts in hopes to provide some insight and hopefully help people with their approach.

In any form of investment one makes in their life, I strongly feel that the investment can not be justified without some degree of quantification that involves variables that have a direct effect on the investment itself (i.e. free cash flow analysis with stock valuation). The better the analyst (or handicapper) can quantify a given investment prospect, the better long term returns he should derive.

As for baseball, each year for over 15 years now, I have developed some sort of power rating system that entails all aspects in a baseball game that has importance, all weighted to what I deem the amount of influence they have on a particular game. Once all these variables are combined for both teams in a particular game, I find the disparity of the intrinsic value of both teams given. For all the games (nearly all) for 15 plus years, the wins and losses for each given disparity has been noted and plotted. Over the years, there have been structural and dynamic changes in the game itself that have effects on the variables used, so keeping the valuation process and ongoing and dynamic as possible is vital. To find the accuracy of the quantification process, their should be a linear relationship between the wining percentage and intrinsic value disparities (power ratings) of particular teams, and fortunately for my valuation process, any given 4 year time span there has been a perfectly linear relationship ( give or take a couple small basis points). For example, if teams that I have quantified to have a 40 intrinsic value has winning percentage of 53.5%, then an intrinsic value rating of 45 should have a winning percentage greater than 53.5% in the long run.

Some of the variables that I use in my valuation process for baseball are the following.

Starting pitching: A pitchers pitching performance is never the same, and his expected performance can be quantified to a certain degree, hence, his intrinsic value should not be the same every game. Things I look for in deciphering the intrinsic value of a pitcher on a given game is his current form, how his style of pitching matches up with the lineup he is up against, how his style of pitching matches up with the ball park he is up against, past history against batters in the expected lineup, how his style of pitching matches up with the strike zone of the umpire, if the pitching has a strong history of favoring home or a way, and several more fundamental variables that effect his expected outcome. Then there is a technical standpoint in the valuation of a pitcher as well, and that is his pitching chart. Does his past history show that he is due for a breakout, bounce back start, regression, or does his current chart resemble past charts that show he has the propensity to be a “falling knife” for long periods of time. All these factors are quantified and added are subtracted accordingly to the base intrinsic value of a particular pitcher

I have seen many fundamental models over the years, and one of the more common mistakes I have seen in a handicapper’s valuation process is the propensity to double count starting pitchers. This implies that they derive and intrinsic value for the entire team that includes the starting pitching, and then “double count” the starting pitching in the equation by adding the worth of a starting pitcher on a particular game, thus making their model obsolete.

Hitting: Deriving the intrinsic value of a teams hitting on a given game is quantified on a player by player basis. This intrinsic value of a hitter changes less than a starting pitchers on a game by game basis, does not factor in how he matches up against a starting pitcher (as that would be double counting as well, since that is done in the valuation of the starting pitcher). However, some of the variables that do effect the intrinsic value of a particular lineup on game by game basis is how well their hitting for is of late (on an aggregate level) how their style of play is correlated to the ballpark, whether they hit better at home or the road, “the umpire effect” (large strike zones diminish the worth of better lineups than less potent ones), and a few more variables. Note: the original intrinsic value is based on expected influential fundamental stats and then quantified into a power rating.

Bullpen: The intrinsic value of a bullpen is derived on a player by player basis that it based on key expected stats that are then quantified into a power rating. It is then weighted by the importance of each players role in the bullpen, and altered on a game by game basis based on key variables that effect the expected worth of a bullpen ( how much have the worked of late, how they match up against the opponents lineup) ect.

Then their exists other key variables that have less effect on the outcome of the game and are waited accordingly, such as weather, defense, managing, and other variables out of the realm of this write-up. Note: other key variables don’t take in account for streaks that are often accounted for by other fundamental cappers, as that would be “double counting” as well, as the aggregate change of intrinsic value of hitting and pitching change based on current form, and in my opinion is more accurately quantified rather than going by streaks (that are quantified as a whole and have the propensity to account for a pitcher that had an adverse effect on the streak itself. There are also a few other variables often used by fundamental cappers that are not accounted for due to double counting as well such as home field advantage that is quantified within the intrinsic value of each variable, which in my opinion gives a more direct worth of home field advantage on a team by team and game by game basis than a generic basis point deduction or addition.

There are also many fundamental cappers that rather than take the disparity of intrinsic values and then find the expected winning percentage based on thousands of past games with the same intrinsic value disparity, and then take that winning % and quantify it to a line (i.e. a home teams intrinsic value disparity of 62 over the last 5 years has a winning % of 60 (note an arbitrary example that prevents math). The intrinsic value of the home team would then be -150, and a market price (that accounts for vig) below that can be justified. Rather than use this approach, many other fundamental cappers use a basis point addition and subtraction approach in which the add basis points for a particular pitcher, basis points for home field, hitting and so forth. However, in my opinion, the basis point approach is more arbitrary and harder to quantify, although can reap very successful long term results when done correctly as well.

So that is how I derive a fundamental intrinsic value line. However, not mentioned is the regression analytics that accompany the valuation process (especially later in the season).

The name of the game in handicapping is coming out with a sharper line than the linesmakers, which is obtainable due to the expected supply and demand they have to account for in their valuation process, but has no effect on an outcome of a game. This fundamental approach allows one to grind out long term profits, and with proper money management can reap long term rewards.

To each is own, there are millions of ways to skin a cat.

LargeMouthBass said:

To each is own, there are millions of ways to skin a cat.

Click to expand...

Word life dun.

At the end of the day, it's all about making that paper. Doesn't matter how you do it.

LargeMouthBass said:

To each is own, there are millions of ways to skin a cat.

Click to expand...

Well said.

I only want to add this:

In every sport, there is something about the game (or, rather, IN the game) that can hurt you (the gamer). Identify that first, then avoid it. Learn how to find it, then stay away from it.

Lastly, when you start skinning cats, figure out WHY that particular technique is working for you. Then, stay with it. Don't run off trying to find another way to skin something.

Finally, you must watch the money. Before, during and even after. ML is most important, more so, than a spread affair. If it's a spread-type game? I still look at ML. People downplay this = it is a mistake. Chart it, or set-up a program that does it for you. You will begin to see things in a, ah "clearer" light, then go back and compare (that requires keeping lots of records and a little of your time. It will help you win.

I agree with all said = but I gotta go = I've been skinning the hell outta cats lately. More today.


us in land of ozone and cat-skinners

Dsethi said:

But, and even though I don't consider myself "sharp" or even a good capper, I don't agree that sharps bet on prices alone.

And if you perceive a line to be significantly off, isn't that in itself based on the teams that are playing and your perceptions of them and the matchups involved?

Click to expand...

I don't think he said sharps bet on prices alone, but it's probably the biggest factor (although a strong argument could be made for money management). The O's today is a good example IMO. Do I think they'll win today? Not particularly, but at 2-1, it makes for a solid value. Just like in hold 'em . . . sometimes you call a bet when you know you're behind because the odds dictate.

While you can win any which way over a short span, I think people should be open to those who have done this for years or decades and have seen betting fads come and go. In the end, there really aren't that many ways to win over the long-term, despite what we may see here over half a season or a season. Just my $.02.

nice read BG (thanks) but it says nothing for the most important aspect in betting and that is (money management)...... In staying focused and disciplined and stay "off tilt".....

there are plenty of people and i know them personally, that can bang out 55-56% winners (long term)(against a -1.10) but never win money......... why, because they lack the discipline and the money management skills to ever become a winner......

Ive made this statement many times on this forum. Its (at least) just as important "How you bet" as "who you bet".

thanks for the post....

dsethi, you are well on your way....... now dont let them put you on tilt....

Dsethi said:

last year/2 years ago I bet more on "value" that it seems you're referring to. If Baltimore was getting a great price that I felt was way too high, I'd hit it. Terrible results for me. This year, if I feel a team will actually win the game or has a great chance to, then I'll make a play, but I've cut out just playing games based on the discrepancy between what the true line is and what I initially perceived it should have been.

Adds to my attempt to be more selective, I suppose.

Click to expand...

All I'm saying is that, like in poker, we can't be too results-oriented. If I think a team will win, but they're at -200, I usually won't play them. If I think they have a 3-1 shot at winning though, it's clearly a good value. Just like when I push with my set on the turn, and player x turns over a flush draw which hits on the river, I chalk it up to a play I'd still make every time, even though it lost this time.

Dsethi said:

I'd never play a -200 line regardless...

Ex) Texas was a large dog yesterday to Oakland, and it seemed like some cappers were on the Rangers b/c they've seen Haren quite a bit before and Oakland's offense is woefully mediocre. I didn't touch that game, and if I did it would have been on Oakland at -1 or -1.5, b/c realistically I don't think Texas wins more than 3 times out of every 10 they face Haren, if that.

I don't know, I'm just done with "value" plays, I'd rather play teams at a low or + juice that I strongly feel will get a W that night.

Click to expand...

That's the exact example I'm talking about. You determined your own values of what percentage of games Texas would actually win, and used that to determine which way a play would go. You really wouldn't lay -200 on a team even if you strongly believed that the team would win 3 out of 4, or more?

As another example, would you only play a dog of +200 if you thought the dog was over 50% chance to win (a situation I can't see happening often, if at all)? Or would you play the +200 dog if you determined that they had a 40% chance to win, which statistically would show a profit?

Dsethi said:

last year/2 years ago I bet more on "value" that it seems you're referring to. If Baltimore was getting a great price that I felt was way too high, I'd hit it. Terrible results for me. This year, if I feel a team will actually win the game or has a great chance to, then I'll make a play, but I've cut out just playing games based on the discrepancy between what the true line is and what I initially perceived it should have been.

Adds to my attempt to be more selective, I suppose.

Click to expand...

I agree. The other day Cleveland looked like a great value @ + 185 with Beckett coming off a 2 week layoff. One of the best teams in baseball + 185. However, Boston & Beckett could and did win the game 4-2. Is it a great value if the team losses?

First and foremost, you have to believe the team is more likely then not to win the game.

SSI,
True, I purposely left out money management in this thread, as I wanted to specifically mathematically debunk the myth that handicapping hot streaks is directly correlated to better results, when in reality, it is directly correlated to a more accurate quantification process. In the past, I have stressed the importance of money management, and will continue to claim that it is the single most important variable of producing long term returns.
<?xml:namespace prefix = o ns = "urn:schemas-microsoft-comfficeffice" /><o> </o>
<o> </o>
Dsethi,
I don’t want to tell others how to handicap, especially one who has been doing well. However, I will say that your statements defy mathematical logic, and this approach is the approach of nearly all squares. It is nearly impossible to grind out a long term profit on “gut feel”, as sooner or later, the vig accompanied “I think this team will win” theory will catch up to you. I will continue to say that no bet is truly justified unless a handicapper believes the team he bets on has a greater chance of winning than the money line indicates. The long term winners can consistently quantify an accurate enough percentage to grind out profits on market inefficiencies. A long term winning handicapper that makes his bets on “this team should win approach” are a statistical rarity, even more so than the 2% rarity of successful long term cappers.

Willie99 said:

I agree. The other day Cleveland looked like a great value @ + 185 with Beckett coming off a 2 week layoff. One of the best teams in baseball + 185. However, Boston & Beckett could and did win the game 4-2. Is it a great value if the team losses?

Click to expand...

Yes and no. It all depends on your valuation. My valuation had the Sox -1.5 as a bit better than the Indians money line. But last night I thought the Indians at +165 were the best value in that game, despite the fact that I thought the Sox were rightfully favored.

First and foremost, you have to believe the team is more likely then not to win the game.

Click to expand...

This just isn't the case, just like as in my poker analogy. If you know a team will win 1 out of 3, but yet you are getting 5-1 on your wager, then there is clearly value. I'm no expert, but I do know that setting a requirement that you have to think a team is more than 50% to win is not a good strategy. It leaves you playing almost all favorites (they're favorites for a reason . . . because they will win more often than not). Again, we can't become too focused on results over the short term.

Willie99 said:

I agree. The other day Cleveland looked like a great value @ + 185 with Beckett coming off a 2 week layoff. One of the best teams in baseball + 185. However, Boston & Beckett could and did win the game 4-2. Is it a great value if the team losses?

First and foremost, you have to believe the team is more likely then not to win the game.

Click to expand...

I cringe every time I hear such a statement. Value is derived before the game takes placed. It is as simple as that. The accuracy of the value quantified is analyzed over a LARGE SAMPLE SIZE of bets in which the handicapper quantified value- not one bet. I guarantee you that any sharp bettor that quantified value on the Indians that game, quantified a value in which still gave the Red Sox a greater chance of winning. The fact that the Red Sox one that given game does not mean they had value. Analyzing where money line value lies on a post hoc basis is simply wrong. You can not look analyze value on a game by game basis and dismiss a bet based on the notion you bet on value even if the game lost. This logical thinking is a thought process a lot of people quickly dismiss as they look for the simple way out. <?xml:namespace prefix = o ns = "urn:schemas-microsoft-comfficeffice" /><o></o>

Willie99 said:

I agree. The other day Cleveland looked like a great value @ + 185 with Beckett coming off a 2 week layoff. One of the best teams in baseball + 185. However, Boston & Beckett could and did win the game 4-2. Is it a great value if the team losses?

First and foremost, you have to believe the team is more likely then not to win the game.

Click to expand...

I also had Cleveland in that game but i also had them yesterday at +169 and that's why it's value i went 1-1 but still went on top +69