Understanding Variation in Thoroughbred BreedingQuality in Thoroughbred breeding cannot be assured. Defects cannot be eliminated, by simply acquiring expensive mares or breeding to expensive stallions. Though hard to understand, variation uniquely affects Thoroughbred breeding quality. Statistical variation in Thoroughbred breeding is extreme and can be described by studying a few factors and examples.

Understanding Variation and Smart Racetrack WageringRacing is a parimutuel system. That means everybody bets their money, it all goes into a big pool, the racetrack takes about 20% of the pool money out for profit & expenses AND THEN they return the 80% that's left to the bettors. The odds you see that constantly change as people bet, represent the 80% payback. Those odds are constantly recalculated to be sure winning bettors only get 80% of the pool money back and the racetrack always takes their 20%, and they always make a profit!
To beat a parimutuel betting system you must outsmart the rest of the bettors, not the racetrack. You also have to outsmart the other bettors of course by enough to get back the 20% the racetrack takes out. This may sound easy but it is mathematically more difficult than some people think. You see the average bettor will always get 80% back. The odds assure this by constantly changing. So, if you are as smart as the average racetrack bettor you simply will lose 20% of your money over time. This means, if you are an average bettor and you wager a total of $1,000 in a day at the races, you will lose $200 a day. This is not an opinion or concept. It is an absolute mathematical fact. So, let's say your goal is to just break even and you wager $1,000 a day. Guess what? You have to be 25% smarter than the average racetrack bettor. Your $1,000 break even goal divided by the $800 the average racetrack bettor will come away with after the racetrack takes their 20%, result in a 1.25 factor. Thus you have to be 25% smarter. right? Now, let's say you want to be a professional gambler. Your goal is not to break even but you need to cover your expenses and earn a living. If you again wager $1,000 per day, or $5,000 per week assuming you wager 5 days a week, what would you set as your wagering profit margin goal. Let's just say it's 10%. Each day you come to the track with your $1,000 and leave with $1,100, winning $100 a day, or $500 a week, or $26,000 a year. (Of course it doesn't work that way every day but these are averages.) Your simple goal of making a 10% profit margin means you have to be 37.5% smarter than the average racetrack bettor, to earn a 10% margin. (The $1,100 you need to leave with daily divided by the $800 the average bettor leaves with daily equals 1.375, or you need to be 37.5% smarter. With this simple math in mind remember, there are many racetrack bettors who know nothing, but those occasional patrons really do not bet much. The folks who are there everyday, or understand what's going on on the backside (inside information), or gamble professionally, are the big bettors. They really establish the odds you have to beat! Those are the people you have to be 37.5% smarter than to make a 10% profit. So how can you really gain an edge on them? The place to begin to answer this question is by understanding how variation works. You see, all "successful" predictive models begin with the understanding that statistical variation is a key, and PERTINENT DATA can create predictive success. For example, everybody knows that the chance of winning a coin flip is 50%, whether you call heads or tails. This is a known probability but, what if you didn't know this and you were trying to predict the future of coin flips. So let's say you didn't know that heads or tails will come up 50% of the time. What would you do? You would look at past coin flips and try to predict the future, right? If you only analyze 2 coin flips, the chance of getting exactly 50% is pretty rare, correct? On the other hand, if you analyze the 1,000 coin flips, you will be very close to the 50% known predictive value. If you decided to analyze 1,000,000 coin flips you will be just decimals off of 50%, but not really that much closer to the result you got when you analyzed 1,000 flips. (FYI, there is a sweet spot associated with how much, or how little, information is needed to create wagering credibility in predictive modeling.) Predicting the future result of a race is just like predicting the result of a coin flip except the predictive result is unknown. Looking at only the last race, the last finish, the last speed rating, the best race, the last stretch run, the best trainer, the best jockey, or any one data point is as silly as looking at one coin flip to predict the next one. Nevertheless, many bettors at the racetrack consider only a few pieces of "recent" data to wager!, They also have no idea what weight they give different pieces of data from time to time, and constantly vary their predictive methods. In other words, if you know what is important and how much information you need to be reasonably accurate, you can be smarter than the average racetrack bettor! The advantage you can gain by understanding variation when racetrack wagering, or frankly any endeavor which rewards you for accurately predicting the future more effectively than your competitors, can be valuable. You can beat your competitors who use minimal, emotional, incorrect or simplistic information to predict the future. You do not have to be perfect, and no predictive model is. You must only be consistent in your predictive approach over time, and constantly improve your predictive methods by measuring results. Whether your method can generate results 37.5% better than the average bettor, and generate a 10% profit margin, is another story. Nevertheless, you can improve your chances of success by understanding the mathematics of this game, and racetrack wagering is a game where advantages can be gained. Though there is a programmed disadvantage where 20% of the wagered money is raked off by the racetracks, some people learn how to overcome this disadvantage over time. 