Critique My Method

Hi guys, I'm relatively new to handicapping.  My background is in stock picking and I applied a few strategies starting with MLB this year.  I guess it's what you'd call a 'value underdog' strategy where I think the line doesn't reflect the underlying statistics.

I only bet on underdogs where my formula (largely dependent upon comparison of visiting team points scores relative to starting pitcher ERA plus home team points scores relative to visiting starting pitcher ERA) predicts a winner but the line reflects an underdog.

Since the beginning of the season I'm at 52%, maybe not all that sexy, but the avg line is 124, so I'm doing pretty well on the profit side.

I also backtested what I think is called a 'progressive' system or what I would call 'doubling down' and my profitability goes up dramatically if I adjust my bets in the event of a loss (calculating the next bet to break even in the event of a loss, and so on with each subsequent loss including the aggregate successive loss). 

My longest losing streak was 4 days in a row, which happened once, and I've had multiple incidents of 3 losing days, so the most I've ever been upside down was about 250% of my original bet.

If this holds, I'm looking at doubling my bankroll every 45 days or so.  What am I missing? Binomial distribution on the losses?

Well first off, I'm glad we finally have a new member that isn't here to try to promote a scamming website.

Back on topic, hitting 52% on dogs is not bad at all, in fact it's quite good. I could never hit 50% on dogs in baseball, was at 48%, still made profit though. In NHL I like dogs because I can hit 56% :P

Anyway, be careful with the progressive betting as one bad streak can bankrupt you. Hitting 52% with dogs might warrant you to stay away from the chase altogether and just increase your bet size.

With chase systems, they really only help if you're playing the same team (or against the same team) throughout the chase.
Just doubling your bet if you lose won't help too much as you will occasionally lose several times in a row.

If you do decide to stick with progressive betting however, I would suggest only doing a 3 or 4 game chase, 5 at most.

Thanks for the reply.  I've seen both 40% and 44% as the statistical 'upset' figure, do you know which is correct (if any)?  If I can maintain a long-term avg of 52% and either number is correct, that seems 'good' to me, although I don't have much perspective.

Forgive my ignorance on the terminology, but I don't see how a progressive money management method is dependent upon staying with the same team; the probabilities of victory are not altered by the team, but rather what the likelihood of winning is on any particular bet.  If I flipped a quarter 10 times, and then I began flipping a nickel, the odds really are no different.  I'm more comfortable going with the highest probability bet.

I've researched as much as possible on these progressive methods and I don't think most of the promoters really have any idea what they're talking about when it comes to the statistical part of it.  I recognize there is a risk that a pattern could emerge that could wipe you out, but I'm not sure yet how to manage that so I can take advantage of the high frequency of win-lose-win or win-lose-lose-win (both of which occur very often in a 50/50 scenario).

I've got about 100 'events' to study now, so it's not a great universe but it's adequate.  My longest win streak was last week-5 wins in a row (4 on the Marlins!), so I guess that's a reminder that I'm due for a 5 loss streak as well.

Thanks for the feedback and I appreciate any further comments.

For MLB 44% is correct.

The reason following the same team is more effective in progressive betting is the motivation factor.

If you were hitting +100 at 50%, the possible solutions are...
WWW +3u (+1u if chasing)
WWL +1u (+1u if chasing)
WLW +1u (+1u if chasing)
LWW +1u (+1u if chasing)
LLW -1u (+1u if chasing)
LWL -1u (+1u if chasing)
WLL -1u (+1u if chasing)
LLL - 3u (-7u if chasing)

Since each has an equal chance then in the long run you'd break even, just as you would betting each single game. (12-12)

The following is hitting 50% at +120

WWW +3.6u (+1.2u if chasing)
WWL +1.4u (+1.2u if chasing)
WLW +1.4u (+1.2u if chasing)
LWW +1.4u (+1.4u if chasing)
LLW -0.8u (+1.8u if chasing)
LWL -0.8u (+1.4u if chasing)
WLL -0.8u (+1.2u if chasing)
LLL -3u (-7u if chasing)

So betting a +120line at 50% using straight bets would give you +2.4u.
Chasing would give also give you +2.4u.

As you can see, losing 3 games in a row does so much damage that it would make it no different than betting each game on its own.

Tomspeople:  If your assertion as to motivation is true, then it should be measurable, as in:  "If a team has lost 3 games in a row, they are xx% more likely to win that that same team is likely to win if they have only lost 2 games in a row".  I'm still inclined to favor the best overall win as opposed to the team on a losing streak; after all, maybe they lost 3 games in a row at home, and now that same team is on the road against a stronger opponent. Wouldn't I be better off moving to another team that is a better statistical opportunity?



Maybe I was imprecise;  after the first loss, I adjust the subsequent bet only to break even, so if the loss is -1u and the line is 1.2, the next bet is .833u.  At the next win, we achieve break even, and the bet level returns to 1u.  Thus far my longest losing streak was LLLL and the bet amt was 1u, .9u, 1.55u, 2.5u, 4.25u at which point the 5th play (4.25u) resulted in a win, and brought me back to break even.  The next bet was of course only 1u, and if that is a win, then the next bet is only 1u.

I admit I am applying strategies which have worked for me in the stock market in a very different environment, where there is no real time risk associated with the use of additional capital to follow a 'loss', whereas here the investment grows with each subsequent loss, but still, in the example above we go from only 1 unit to a final use of 4.25 units and a 'drawdown' of a total of 9 units.  The risk seems reasonable considering the impact to the bottom line profitability.

Whatever works for you I guess :P

Anyway, hitting 52% on just dogs is great so keep up the good work :)

Also, just curious, what is your unit size?

Tomspeople:  I'm not here to advance an agenda, I'm looking for feedback!   But I admit a preference for quantifiable processes!

My unit size is only $500.  I chose that because I had a comfort level knowing I might end up with 5 unit bets and I didn't want an emotional resistance to sticking with the strategy. I guess coming from the stock world and big trades has made it easier than if I didn't have the experience.

Firstly you are using proressive staking, try something else.

Secondly, I don't think with 4 days will are even close to your max. drawdown or longest run of outs.

Whats your sample size? Min 1000 instances before you can begin to determine if you have a profitable system or not.

Try something else? As in...?

Obviously 4 days is not the maximum, it's just the worst experience in the 100 or so days.  As I mentioned above, I've already had a win streak of 5 days which serves as a reminder that a 5-day loss is due.

This is really about binomial distribution, so I probably need to develop a rule as to how to handle a long-streak against me.  In stocks I use a 'stop loss' that protects my down side, so I need one here. 

There's an interesting article on binomial distribution on this guy's site:  https://www.professionalgambler.com/binomial.html

I guess right now my odds of say, a 5 game losing streak are about 3% and the odds of a 7 game losing streak are about 1/2 of 1%.  Sound about right?

This is where I come in, (University for math and statistics :) )
At 52% the of getting the follow losing streaks are:
1 - 48.00%
2 - 23.04%
3 - 11.06%
4 - 5.31%
5 - 2.55%
6 - 1.22%
7 - 0.59%

Now the chance of losing 7 times in a row may be low, but if your on a 7th game of a chase it will cost you almost 130units if you lose

Tomspeople: I don't see where you get the 130units; after 4 losses I was at 5u bet and down a total of 9; within two more bets I'd only be at 30u down (still enough to be worrisome).

So how do people manage the risk?  Is the progressive style simply unsupportable long-term? Or do you employ an arbitrary rule, such as stop after 5, regardless.

All the math is great but I think the K.I.S.S. applies here.   After more time has passed the bottom line is that if you can keep hitting 52% and an average line of +120....you will make lots of money.  Period.  Whether people believe you can or not doesn't matter.  If you have a method to produce similar results of your first 100 days just stick with it and take it to the bank.

You would double up after each loss.
1
2
4
8
16
32
64

The total is 127.

Tomspeople:  As a matter of clarification, I'm not doubling up.  As I mentioned in an earlier post, after the first loss I'm only betting enough on the subsequent loss to break even (cumulatively).  So after a 1u loss and a line of 120, I'm only betting .83u, and so on. 

I thought the doubling up was too aggressive, and made a 5 game streak far too risky.  Although my money management is more conservative, the idea is merely to reduce risk rather than increase profit.

I understand what you are doing now. You mention progressive so I assumed.

Don't get me wrong, I'm here for feedback; after three months of almost doubling my bankroll monthly, I'm looking for cracks in the system before I celebrate! In 10 years of trading equities I never made this kind of money.

As an example, I bet FLA over LOS @ 140 on 7/13 and lost.  Last night I took PIT over COL @ 141 (bet .7u) and lost, so tomorrow, assuming another 140 ML, I'll bet 1.2u (to break even).

Hey guys, I had two losses in a row (FLA on Wed and then PIT yesterday), today I'm going with COL over SEA and I got a line of only 122, so I'm in for 1.45u.  I'll let you know how it goes.

CLE was a loser for me last night so I'm 3 games down.  Today I'm going with LAA over BOS.

dharbour: I'm sorry but I don't understand the system you described.  If you read my initial post above you'll see I only bet underdogs and following a loss I only bet enough to break even, so even with a streak of 3 losses I was only betting about 3units by the final game (which in this case was LAA and we won). 

I've been working on a formula that considers several factors, including:  home team ppg vs visiting team era, visiting team ppg vs home team era, plus the same formula which considers the starting pitcher ERA instead of team, and finally, i have a formula which evaluates what I call the 'line value', i.e., the value of the ML vs my own probability of a win by the underdog.

In other words, I'm looking for underdogs I think are basically a 50 percent chance to win but have a big ML, and since the season began I'm about 53% winners with avg ML of 124.  (Someone above said underdogs are 44% winners, so I'm only 9 pts better than a monkey). 

I'm also using what I think is modified version of a 'progressive' system, so I've doubled my bankroll each month thus far.  In the market I only did 30% a year so I'm thrilled!

I'm sure someone out there is alot smarter than I am, which is why I posted.  At the moment I'm most concerned about my risk management method, although I can't say I've gotten any input on how to improve it.  I'm also not giving ANY attention at all to anything other than these 'technicals', so I'm probably making some bad picks because I'm not even checking to see if a key player is hurt, etc.

I should have added that today I went with MIL over SFO. 

dharbour: thanks for the idea on the bullpen era.  About the trend analysis though, I'm a skeptic on that stuff b/c I've never seen any evidence that a team's wins in the previous 10 games are a predictor of their success in the next game.  That's my hesitation; my current input data is all directly related to the outcome of a game, i.e., whoever scores most wins.

Another practical concern I have is that I don't even have a play every day as it is.  Sometimes my formula just doesn't give me anything to work with on that day, and although I'd rather play more times even if my win rate drops, I'm concerned with fixing something that ain't broke. 

The idea of playing teams with so many wins in their last ten does help, mainly because baseball is a game of streaks. You'll soon learn that if you study it more, that when a team is hot, they stay hot for a while, and vice versa.