Seven Deadly DATA Sins in A Garden Of Eden

14518083760_c089348f3e_zWhen people think of the 7 Deadly Sins, they think of Adam and Eve, the apple and the snake. But, as with data, the landscape of those sins is constantly changing. What used to be a best practice is now a haunting memory of failed strategies. What works for one sample can prove disastrous for another. It’s our job as analysts to avoid these pitfalls. So here’s my take on the Sins, and how you can stay in a garden of Eden for data.

 

Don’t be Greedy. You have a great model. It fits perfectly. The sample offers positive results. You are almost sure of your path. Then you think, maybe I should add one more variable….DON’T. Overfitting is one of the easiest ways to take a winning algorithm and let it take the plunge. Be meticulous in your work, but know when to stop.

Which leads us to: Gluttony. Your model might be perfect for futures contracts, but work out awfully in housing markets. It’s your job not only to know your strategies, but where they work best. Don’t get drawn in by positive results and think you’ve found the panacea of data. You haven’t.

Once you have your model, it’s time to look at the data. Here is the spot to be wary of Lust. Out-of-sample data is so useful. But don’t let the possibilities that lie out of your sample distract you from the in-sample data. You want the best, most accurate results for your model.

For accuracy, avoid Sloth at all costs and as I said before, BE METICULOUS. When creating your sample, go the extra mile. Search through historical data. Look at the algorithms of other traders and analysts. Check the credentials of your data sources. Nothing is as toxic as bad data. You can do everything right, but based on the wrong data, have disastrous results.

Test your strategy in the real world. Celebrate if things go your way. But if they don’t, don’t let Anger cloud your judgment. If your model doesn’t fit, no amount of anger will fix it. What will is looking critically at your model and assessing WHY it didn’t work. When you step back from your emotions and use your analytical knowledge, that is when you get your best work done.

It is great to admire other analysts. It is fantastic to learn from them. It is dangerous to Envy them. When you put the strategies of someone else on a pedestal, you forget your own expertise. Trust your knowledge. Trying to copy someone whose ideas differ from your own gut instincts will be disappointing.

Finally, like I said, have faith in your expertise. But as with everything else, keep it in moderation, and don’t let Pride overwhelm you. No strategy is going to work every time. And no data sample can truly predict the future. You work in the scope of probability. Play the game, but always keep an eye on where you stand.

 

Quandl.com

 

 

1 Response

  1. Heck yeah this is extcaly what I needed.

Leave a comment