Lessons from Moneyball
37signals.com
37signals.com
This is the same strategy that Warren Buffet has been using for decades.
The perception is high risk because you are going against the status quo, but in reality you are in a low risk senario because you are doing what you love and you see an undervalued asset.
In the startup world it is the same thing. If you are doing what you truly love/what you are good at and you can identify undervalued assets then you will win in the long term.
For employees this is finding a company that you can calculate that it will succeed, then getting stock in that company by "taking the risk" and it all works because it is something you deeply care about.
For founders this is finding an area that is ripe for disruption with a problem that you face in your life. Starting a company is a way to realize that value that other's are not capturing.
Really successful companies, founders, employees and investors do what they love, love the ones around them, pay respect to everyone and can brutally identify undervalued assets.
No, it is high-risk. The risk is that the undervalued asset you see may not really be one.
But staying with the status quo is high-risk as well.
Risk is a red herring; people optimize for regret-minimization. There's less regret if you go with the crowd because there's other people to blame -- and regrets are less apparent because everyone suffers together. But if you go against the status quo and your life sucks, you have nobody else to blame, no fellow sufferers.
This isn't a rational thought process. But you have to see the enemy clearly before it can be defeated.
> This is the same strategy that Warren Buffet has been using for decades.
Buffet has been using govt and others to produce undervalued assets that he then buys. (Did your CitiGroup investment have a govt guarantee?)
He also pushes tax policy that directly benefits him. (Hint - the "millionaire's tax" that he's pushing doesn't apply to him. Same with the estate tax - virtually none of his estate will be taxed. He sells insurance to other folks who will have to payit.)
Guess who benefits from delaying/killing Keystone XL? (Buffet's railroads aren't nearly as efficient as pipelines.)
1. Always question the conventional wisdom. That's what needs to be disrupted.
2. Every great innovation that makes someone famous or rich always involves some guy who's been plugging away for years on his own without any notoriety or wealth (In this case it was Bill James, who invented "sabermetrics" while working as a night watchman).
For web software development there isn't even a clear way to determine winners - e.g. market share or brand awareness may trump revenue as a measure of success depending on the purpose for which web development is undertaken.
If Bean's goal had been to put butts in seats, it is unlikely that his methods would have worked.
However, it is unlikely that a strategy optimizing dollars/win is the same as dollars/season. Merchandising and TV contracts play too big a role (not to mention ticket prices). Stars sell merchandise and seats even for losing teams.
http://onstartups.com/tabid/3339/bid/76799/Startup-Lessons-F...
What I find most interesting about the story is the different effect it has on people. Business types see it as "Take a great risk and it'll pay off (literally)." Science types see it as "Math works".
The reality is, you just have to do what you want to do and not care if other people tell you you're crazy. Was a situation like Moneyball an exception to the rule? Hell yes. But with out those 1 in 1000 exceptions to the rule we'd have very little to show for 50 thousand years on the planet.
Do not listen to detractors IF YOU HAVE EVIDENCE SUPPORTING YOUR CAUSE, but do not blindly pursue something because of an unsupportable "gut" feeling
I thought the real lesson was "massively outperform the limits of your shitty resources beyond anyone's expectations, by thinking outside the box and concentrating on algorithms and data modeling; and still not succeed. Then if someone is impressed and offers you more personal compensation than you can imagine, to do the same thing but for them, while actually being given enough resources to actually have a chance with it, then you go and fucking do it."
The startup analogy is easy: if you are solely responsible for running a free service with a million users on $300/month hosting budget, and you almost do it by inventing amazing algorithms, but the service still lags and sucks, but is almost there, then when Amazon makes you an offer to come and do the same thing for them you fucking take it.
He was rejecting the hoity-toity New Englanders and their fancy tea served in fancy porcelain teacups. However yes I'd agree that wasn't sufficient explanation. Maybe there are more details in the book?
I'm still reading through, but, you have two major actors in your brain. A "fast" side, and a "slow" side. The fast side makes gut decisions, and the slow side, well, applies more complex judgements and can include a more unbiased view of the world. There's a ton to this process, but importantly, the fast side is very happy to jump to conclusions when presented with missing information.
The lesson of Moneyball, then, was that the "fast side" thinking clouded judgements of baseball prospects by filling out ridiculous statements like "he's got a hot girlfriend, must be able to have the confidence we need on the field".
I'm not sure how you can use the baseball metaphor in terms of software design, though.
But, you could easily say that we're presented with an incomplete amount of information when building software. (What's going to happen when it runs? Who knows! Just build a nice API and continue.) So we definitely all probably have a lot of "quick learning" techniques and processes we just adopt to. For example, Googling before you just stop and think. Or using the UI techniques of other applications, because "that's how the big profit-making companies do it".
Teams often contain players whose value is off-pitch or subtle: leaders, morale-boosters, supporters.
The system that Moneyball demonstrates would struggle to find the a manager that would uses a Moneyball-type system. Where are the stats for the team-picker?
While I'm definitely not saying that there is magic in sports (nor any successful website that ignores A/B testing) I am saying that sometimes a talented spotter (or a talented designer) can produce a fit that a spreadsheet cannot.
(yet.)
(Conformist... Yeah... http://www.youtube.com/watch?v=V4CggHUypjY)
EDIT: I guess I'm just allergic to cults.
This is true of any popular figure/entity that has earned some authority in an area. PG, for example, could bust in here and tell us the world is flat and a decent amount of people will support that statement (nothing against him, just a hypothetical example). Now, I really really respect 37Signals, I adored Rework, and a lot of their posts are really great but lately I've started to notice the same thing as you, pron. Not every post of theirs is a gem worthy f this front page and I've started to notice a smugness in them that I don't really like but I'm just one guy who has not earned the same respect and authority as them so I suppose I don't have much room to talk here.
Why do posts make it to the front page? Because enough people upvote them, so there must be something that appeals to at least some people. If you have comments on how I could have made this post more interesting or relevant to you, or improvements in my writing style, I'd certainly appreciate hearing them. You can reach me at noah@37signals.com.
Now, what I was talking about is a pattern I see where posts from particular sources end up here very frequently. When I see that happen I start to wonder if they're here because they truly are great articles or just because they came from a particular source with a lot of clout. There's nothing wrong with this in and of itself, it just makes me wonder if we're starting to give particular sources a free pass, allowing them to ride the wave of their previous successes without really evaluating if the latest article truly is worthy of so much recognition. For me, the test would be to put the same article on an unknown blog and see if people still feel the same way about it.
Please know, that there's nothing about the content or style of your post that I object to at all. My mind just goes off in that direction after seeing any source on the front page very frequently in a short period of time. And you're right that this on the front page because the majority found it appealing. We have a great community here and I trust them to choose what's good. I want to thank you for posting this comment here because it forced me to be honest with myself and ask, "what made me react so negatively/too critically to this post being here?" and while the reason I mentioned earlier is still true I have to admit that there are two less than desirable traits or attitudes that I let slip out.
The first was plain old jealousy. The "I wish I was successful and cool like 37Signals" type jealousy. The second was negativity for negativity's sake. It was like I just had to go against the grain here for the sake of going against the grain. I'm also in a bad mood today which doesn't help. I think I have to admit I was partly just being petty and that's not cool. Rather than delete my previous comment I'll just man up and admit I was letting my inner asshole/troll slip out and maybe others will be able to learn from me and catch themselves when theirs slip out. Anyway, it wasn't a bad post at all and I'm surprised you even took the time to respond to this unwarranted negativity.
In addition (once on the front page) there's this whole cult following thing. E.g., anything about GitHub is certain to get a whole-lotta love.
In addition, I think (though I may be wrong) the HN algorithm makes second chances nearly impossible. If something doesn't make it to the front page within a couple of hours since its posting (or even 30 minutes), it never will.
The startup corollaries: small-bore inefficiencies can't overcome luck and strong competition, don't attribute success to A when it's really the obvious B (i.e. what differentiates your product doesn't mean it's actually better - Gowalla and foursquare is my favorite example here), and people will steal your landing pages because they work better.
What is this supposed to mean? The season gives you a 162-game sample. The playoffs are a coin-flip, by comparison. It's just too small of a sample to draw any conclusions, which is why both old- and new-school baseball guys have an endless array of cliches about how anybody who makes it in to the playoffs has a chance to win.
In short, the A's roster was actually constructed to succeed over 162 games but the compromises he had to make on defense and power pitching to get extra value are particularly fraught in the playoffs, where you're battling the better offensive teams in the league. Of course, the playoffs are a crapshoot (the Cardinals had no business being in, let alone winning in 2011), but the compromises Beane had a direct effect on their heartbreaking playoff losses in the early 2000s.
The reason the A's didn't win the World Series is the same reason the Cardinals won the World Series in 2011 and 2006, but not 2004. It's the reason the Giants won it in 2010. The playoffs are a crapshoot. It has nothing to do with whether or not sabermetric analysis is valid for regular season games, but not for playoff games.
The whole point of the book was that winning games was done by getting on base and the people who get on base more were better to have on a team than someone who was good at fielding (which the book said didn't matter).
The reason the playoffs were such a crapshoot wasn't because Oakland had a worse defense or pitchers but because the amount of games being played was so small that the results could go either way. as the teams in the playoffs were generally of similar quality.
Beane and the A's were mostly focusing on offense, because that's were they could find value. It's significantly harder to find value in pitchers, and pitchers can help damper some of the "luck" factors in the postseason, though it's very much open to randomness.
Modern baseball management is focused on building your team to win enough regular seasons to make the postseason, and then hope luck falls in your direction. If you can couple that with great starting pitching, you can reduce the amount of luck you need.
I'm not sure how relevant the techniques in Moneyball still are. At this point, every team is looking at Sabermetrics, and some of the high-value stuff isn't there any more. For example, getting on base isn't worth as many runs in 2011 as it was in 2002 because there are less home runs hit.
This is not true. One, the big three of Zito, Hudson, and Mulder is one of the best front ends to a pitching staff in baseball history. Moreover, Isringhausen, Bradford, Mecir, and others on that staff were all players undervalued by the traditional ways of evaluating players. They had a team ERA+ 122. That is not a team which is "mostly focussing on offense."
Zito, Hudson, and Mulder were all draft picks. I think Moneyball did discuss the A's strategy to draft college pitchers instead of high school, as Beane found high school pitchers to be over-valued.
Maybe you're arguing that the amount of luck influencing the outcome of the playoffs is much higher than the amount of luck influencing the regular season; this is obvious. What it's not is a good argument for not using all available tools to put the best talent on the field.
For a playoff, things are different... you're in a match-up against a single team that is better than the average team. I would argue that money and luck plays a much bigger role in the playoffs for a smaller team. If you're playing the Red Sox or Yankees, chances are that all of the key players qualify as "best talent", while your budget forced you to choose your investments more carefully or choose players with less-known ability.
Luck plays a much bigger role for all teams in the playoffs. That's the nature of the small sample size. I'm not convinced that money plays a bigger role in the playoffs for a given team than in the regular season.
One valid consideration, in the context of the 2002 A's, could be that they built their success by adding wins on the margins. Put another way, the 2002 A's on average were one of the best teams in baseball, but the playoffs put greater emphasis on the "top" of the team, or the best players. That A's team was short on all-star talent, but made up for it by adding wins at the margins. But adding 2 wins by intelligently selecting a fifth starter doesn't matter in the playoffs, because said fifth starter will pitch out of the bullpen, if at all.
However, baseball is radically different than it was in 2002 and I don't think that still applies to the same extent. Almost every team is using the kind of player evaluation that Beane was using in 2002. There's still some fogey Joe Morgan types that just happen to be running baseball teams now, but they are by far in the minority. So now you have to be even smarter than Beane was at the time (see Friedman and the Rays) to be competitive on a budget. But the way Friedman et al. are doing it now isn't subject to the same implications as the way Beane did it when RBI was considered a good metric for talent evaluation.
Reading the book, Moneyball, I came away with the conclusion that the success of Beane's system boiled down the fact that they started measuring everything and making empirical decisions based on data rather than emotion or "gut" checks. In turn, this let them find value where others saw failure. In a competitive environment with limited resources (players Beane's instance) this system will work up to the point where few, if any, try it--meaning lots of overlooked value exists.
One point at the end of the book was that the Red Sox, when they hired Theo Eptstein as their GM, brought in one of Beane's assistants to implement the system the A's were using. As more teams start using it, the quantity of undervalued players decreases because everyone starts valuing the same thing.
Start up takeaway: Measure everything and make decisions based on empirical results instead of "gut" checks. That's really what it boils down to. I'm not just talking about the front-end stuff like landing pages, conversion rates, clicks, etc. Measure it all, response times, database operation times, network transmission time, average request size, whatever you can get your hands on. Having all that data will allow you to make informed decisions and identify places for improvement to increase value.
That's my takeaway.
Most people don't lack original ideas, just like there's no shortage of people who would like to be major league baseball players. What people do lack is the ability (or perhaps just the will) to evaluate their ideas objectively.
And while users may evaluate software with their own eyes and emotions, that behavior manifests itself in objective, quantifiable outcomes that have some impact on your business. If measurable user behavior doesn't impact your business, I'd guess you aren't really in a user-driven business.