The 5 Signs Of Terrible 2015 Bracket Picking Advice

posted in NCAA Basketball, NCAA Tournament

It’s 2015 NCAA tournament time, so two things are pretty much certain:

  • You are going to get bombarded with bracket picking advice. By the media, friends, family, and that random sketchy looking dude at the bar.
  • All that advice will suck. In fact, if you listen to these misguided souls, your odds to win your NCAA bracket pool will probably get worse.

We’ve spent over a decade doing objective, data-driven research into bracket pool strategy. So we can distinguish fact from opinion when it comes to bracket picking tips.

Here are the top five things you need to watch out for. (Or if you’d rather just put the best bracket advice to work for you, check out our 2015 bracket picks.)

Forget Any Bracket Picking Advice That:

1. Ignores your scoring system

Bracket pools come in an astounding array of flavors in terms of scoring systems. Some award a fixed number of points per round for each correct pick; some offer a huge reward for picking the NCAA champion correctly; some give big bonuses for picking seed-based upsets.

The exact details of your pool’s scoring system exert a huge influence on optimal picking strategy, and most people don’t pay nearly enough attention to this dynamic. The bottom line is that there is no such thing as a universal “great pick” for all types of bracket pool scoring systems.

Even though it might be way too risky in most pools, picking a 12 seed to make the Elite 8 could be a great decision in others.

2. Ignores the teams your opponents are picking

In bracket pools, there is no prize for getting a certain number of games right. You simply need to get a higher score than everyone else in your pool.

Consequently, the picks your opponents are making have a significant influence on your own odds to win. For instance, if two teams are equally likely to win the 2015 NCAA tournament, but one team is twice as popular a pick among your pool competitors, then you are almost certainly better off picking the less popular of those two teams.

This is a simple mathematical truth, but it’s going to be ignored by almost all of the bracket advice you hear this year. Even if a team is strong, it’s chances may still be highly overrated by the public.

3. Ignores your pool size

Like scoring system, the number of entires in your bracket pool is a key factor impacting bracket pool strategy. As a general rule, the larger your pool, the bigger advantage you will get from avoiding the safest and most popular picks.

The basic concept here is all about risk vs. reward. In big bracket pools, the chances are much greater that a few of your opponents get really lucky with their picks. As a result, if a relatively safe and very popular team (e.g. Kentucky this year) ends up winning the tournament, it will be extremely difficult to win a big pool, even if you did pick Kentucky.

As a result, not picking Kentucky, and hoping they get upset, is probably a much wiser strategy for larger pools.

4. Ignores your payout structure

A often overlooked factor when it comes to bracket picking advice is the payout structure of your pool. In short, having a very steep payout structure (e.g. winner take all) versus a very flat payout structure (e.g. the top 10 scores win a prize) can lead you to significantly different picking approaches.

For instance, if you don’t need to come in first place to win something, and winning any prize would delight you, then you can pick your bracket a bit more conservatively than if you had to come in first place. By doing so, you can actually “tune” your bracket picks to decrease your odds of coming in first place, but increase your odds to finish somewhere in the money.

People who play daily fantasy sports should recognize this concept, as it’s very similar to optimizing your lineups for 50-50s vs. GPP’s.

5. Is based on “Golden Rules”

Any time you hear someone quote a stat like “In the last six NCAA tournaments, at least four teams seeded 10 or worse have made the Sweet 16,” immediately run for the hills. Every NCAA tournament is different, and these types of trends rely on very dangerous assumptions about historical bracket seeding and matchups. (They’re also rarely supported by enough hard data to confidently conclude that they’re predictive.)

In isolation, the fact that a lot of 12 seeds have beaten 5 seeds in the Round of 64 in recent years has little relevance to the 2015 bracket. What if all of this year’s 5 seeds end up being stronger than usual, while all the 12 seeds are weaker than usual? If that happens, then it’s almost certainly not a smart move to pick lots of 12-5 upsets in 2015.

Why Bad Bracket Advice Exists

We hate to be hard on people, because bracket pools (most of them, at least) are first and foremost about fun. But we’re also very serious about winning, since there’s usually money on the line.

The concepts above are verified by actual quantitative research and bracket pool simulations that we’ve conducted, yet an amazing amount of misinformation about bracket picking strategy still endures.

Perhaps not surprisingly, the mainstream media is one of the worst culprits, even as sites like ours have steadily evolved the sophistication of our objective research into bracket picking. So why is that?

At its core, the process of figuring out the best possible picks for a given bracket pool is really just one big, juicy, complex math problem. A lot of great data on teams and pools now exists in the public domain. You just need to gather it and crunch all the numbers in an intelligent way.

As it turns out, most people, journalists and bloggers included, aren’t that quantitative. This isn’t a knock on media folks; some people are great at math, other people are great at writing or talking about sports on TV. But one thing is clear: Watching lots of college hoops or knowing a lot about individual teams do not make one an expert on optimal bracket picking strategy.

To make an analogy: Most people talking about the stock market on TV aren’t the ones making millions trading stocks.

In addition, great bracket strategy is complicated, and really doesn’t translate well to a medium like TV, where short and pithy sound bites rule.

Wrapping It Up

As they used to say in ancient Rome, caveat tournamenti selector (loosely translated, “bracket picker beware”).

This March Madness season, until you hear someone say, “I think that given the current public picking trends, Villanova looks like a team I’d avoid in larger pools with 1-2-3-4-5-6 scoring and steep payout structures,” then it’s a decent bet their advice is worthless, or even harmful.

So when you hear it, simply smile and say, “Awesome, thanks for the tip, bro.” And then check out how we make NCAA bracket picks.

  • AaCBrown

    While I would phrase things more moderately, I agree with the last four points. The first point requires some qualification. The precise rules and payouts are of secondary importance in NCAA bracket pools, most people spend too much time thinking about them, not too little.

    Consider a single game in isolation. Suppose you think the highest seed that can win the bracket will be picked by 70% of the other people in the pool, but that another team has at least an even chance of winning that game. This is a good pick for you, regardless of how many points you get for a correct prediction, or how the other games are scored. A 50% chance of getting some edge on 70% of the pool has to increase your chance of winning. The size of that edge relative to the total points from other games doesn’t matter.

    The pool rules start to matter on your marginal picks. But the relation is complex. If the scoring system rewards upsets, it makes upset picks more attractive. But it also makes them more attractive for other people. If they over-react, as they often do, you’re better off going with more favorites.

    Therefore I think the best strategy is to start bottom up and find picks with good probabilities of success relative to the number of other people likely to pick them. Only with the final close-call decisions is the scoring system a key consideration.

  • David Hess

    I think we’re pretty close to being on the same page here. I agree 100% that the foundation of a good pick strategy, regardless of pool size and scoring system, is going to be finding undervalued teams. However, the size and system do make a difference in which of those teams you should actually go to battle with.

    Let me give you a couple of examples of what I mean.

    ====Pool Size====

    Let’s say there’s a team with a 5% chance to win the title, and only 1% of the public is picking them.

    In a 1000-person pool with 1-2-4-8-16-32 scoring, they’d be a good pick … (5% chance of a correct pick) / (avg 10 people in the pool picked them) = ~0.5% chance winning the pool, assuming one of the people who picked the champ will win, and all have roughly the same chance. So you’ve bumped your “naive” 1 in 1000 win odds up to about 1 in 200.

    In a 10-person pool, they’re not such a hot pick. You’ll almost surely be the only person picking them, but you’ve got only a 5% chance of them actually winning the pool. There’s also some chance that nobody picks the winner, and the early rounds decide it, but that may only gets you back up to around your original 10% win odds. So you didn’t really gain anything from this pick.

    ====Scoring System====

    Let’s say you have 1-2-4-8-16-32 scoring PLUS a seed difference bonus. So in a 5 vs. 12 game, you get 1 point for picking the 5 correctly, and 8 points for picking the 12 correctly. This year, you give each 12 seed a 33% chance to win. Further, let’s assume that 75% of your opponents are picking the 12 seed upset in each game, and that you want to pick the 4’s to advance to the Sweet 16 anyway, so the games don’t matter much for later rounds.

    If you pick all the 5’s…

    — 20% of the time all the 5’s win, and you gain ~3.75 pts on the your average opponent
    — 40% of the time there is 1 upset, and the average opponent gains ~3 points on you
    — 29% of the time there are 2 upsets and the average opponent gains ~9.75 points on you
    — 10% of the time there are 3 upsets and the average opponent gains ~17.5 points on you
    — 1% of the time, there are 4 upsets, and the average opponent gains ~23.25 points on you

    If you pick all the 12’s…

    — 20% of the time all the 5’s win, and the average opponent gains ~0.25 pts on you
    — 40% of the time there is 1 upset, and you gain ~2 pts on the your average opponent
    — 29% of the time there are 2 upsets and you gain ~4.25 pts on the your average opponent
    — 10% of the time there are 3 upsets and you gain ~6.5 pts on the your average opponent
    — 1% of the time, there are 4 upsets, and you gain ~8.75 pts on the your average opponent

    The takeaway here is that even though most of your opponents are picking the underdog, the 12-seed is still the “conservative” pick — i.e. they’ll maximize your average expected score both in absolute terms, and relative to your average opponent. And picking the 12 maximizes the percent of the time that you gain ground on most opponents.

  • AaCBrown

    I agree we mostly agree.

    While both your examples are true as far as they go, they leave out an important counterbalancing factor.

    In the first one, in the 1,000 person pool, it’s rational for people to pick more upsets. Therefore, I think the example is more realistic if a higher percentage of people in the 1,000 person pool than the 10 person pool pick the 5% longshot winner.

    In the second one, the reason 75% of people are picking the 12 seeds is the attractiveness of the seed difference bonus.

    I don’t claim things offset perfectly so you can ignore the scoring system. For one thing, people don’t play perfectly. For another, even if they did play perfectly, there are stupid plays you could make (like picking that 5%-chance winner in a 10 person pool).

    What I do claim is that other players react to incentives, with reduces the value of adjusting your selections to the specific rule set; and often people over-react, so you should adjust in the opposite of the obvious direction.

    I don’t think either expected value or even expected difference from the average is a good criterion for picks. The really essential thing in any reasonably large pool is to get variance.

    Take any individual game. The selection will only matter if you are tied with the best other entry in all other games, or somewhat ahead or somewhat behind (how much “somewhat” is will depend on the points available for this game). You are much more likely to be behind than ahead, because there are a lot of other players. Therefore, you would rather have a different pick than the other player. To a first approximation, the value of the game to you is directly related to the probability your selection wins, times one minus the probability that a random other player made the same pick. So if team A has a 40% chance of winning the slot, it’s a good pick if only 30% of the field will pick team A, but a bad pick if 50% of the rest of the field will pick team A. It’s like pari-mutuel horse race betting.

    That’s only a first approximation, but it holds true regardless of the scoring system. You generally want to select the teams whose probability of winning exceeds the fraction of other players who will pick them by a lot; and that will be reasonably good advice most of the time.

  • David Hess

    “What I do claim is that other players react to incentives, with reduces the value of adjusting your selections to the specific rule set”

    You will be happy to hear that in our pool simulations, we do adjust the opponent picking rates based on the pool rules. In other words, we don’t just use the national pick data in all pools, we force our simulated players to shift their picking strategies towards the ideal “maximum average expected points” picks. But you make a good point about the overreaction of some opponents. We’ll have to give this some thought in the offseason. Thanks.

    “The really essential thing in any reasonably large pool is to get variance.”

    I agree you need some picks that are intended to provide variance. However, there can be a backbone of “conservative” picks as well. If you’re in a 1000-person pool, and you pick an unlikely pair of teams in the final, and that gambit hits, then you’re either the only one who picked it (in which case the ideal picks for the rest of the bracket would be the popular teams, to provide your opponents less chance to catch up), or you’re one of only a few (in which case the ideal picks would be the safest ones, in order to provide the one or two people that are tied with you a chance to shoot themselves in the foot with the several other upsets they’ve chosen). If the gambit misses, well, you aren’t winning the pool anyway in that case.

    “You generally want to select the teams whose probability of winning exceeds the fraction of other players who will pick them by a lot; and that will be reasonably good advice most of the time.”


  • Groucho

    You’re both wrong – you should always pick the team with the prettier jersey.

  • StevenB

    I think that’s what they both said!!!

  • disqus_yUxiZXaN2b

    No. The team with the prettier jersey won 62% last year, but 91% of the secretaries picked them.~


    Haha, nice one!

  • Gary D. Batini

    My favorite time of the year is here, the NCAA Tournament. You guys have made me a lot of money in the past and I can’t wait for your analysis on the NCAA Tournament this year.

    From what I’m seeing this year, it looks like it is Kentucky and everyone else looking to get to the title run to challenge the Wildcats. I have two early questions to throw out there right now.

    1. Do you look at have a majority of your brackets with Kentucky winning the tournament assuming that is going to be the pick that a majority of the brackets will have as their champion? In that situation, you are looking for the team that will challenge Kentucky in the National Championship game.

    2. Do you try to find the team that could beat Kentucky in the Tournament since it has been around 40 years since the last time a Men’s Basketball team went undefeated and won the title? If you look for the team that could beat Kentucky, who would be the teams to focus on in the tournament?

    I’m determine to have a bounce back year this year after having an off year last year. Any information you have will be good for me this year. Let’s have a great year and win multiple pools.