I'm no stock expert, but

[89th]

It seems one day the market is down, the next day it is up. I remember in college I would always pass a sign at a bank that said what the S&P was for the day. Every day it was the opposite of the previous day. I'm no expert, but I'm pretty sure this is a foolproof strategy... any day the market is down, put a lot of money for the next day and you'll be sure to make a profit, then withdraw before the market closes and sit tight for the following (down) day.

Man, I should open an investment firm.

[89th]

Now I want to download historical data from the last 20 years or so and see how often this happens and then run some scenarios on it. Who wants to be my first client?

[Mik]

[Klaus]

@89th said in I'm no stock expert, but:

Now I want to download historical data from the last 20 years or so and see how often this happens and then run some scenarios on it. Who wants to be my first client?

I've been down that route quite a bit.

There are various places where you can download historical data, but you usually only get one number per day. If you want a finer resolution (say, a price point for every minute) then it gets more expensive.

I used to run a data grabber for a couple of months that collected the data I wanted in minute intervals. At the time, I had an idea for a novel pair trading strategy, which did actually work when not considering transaction costs, but the transaction costs at the volumes I could afford made that infeasible. My idea was an analysis to find temporary cointegration between two assets - an inverse correlation that would only last a couple of hours or minutes - and make use of it while it lasts.

[jon-nyc]

There’s a whole branch of HFT that does very complex versions of this. Imagine taking it many levels deeper - do movements in Intel at all predict or can they be predicted by moves in MSFT? That’s a trivial and dated example.

These guys have databases at the tick level (not just daily close) which they constantly scan for correlations, most of which wouldn’t be obvious using a more fundamental approach. Also it’s done cross asset class. (Eg relationships between currency exchange rates, interest rate futures, and stock prices)

This is sometimes referred to as “statistical arbitrage” or “stat arb”

[Klaus]

@jon-nyc said in I'm no stock expert, but:

There’s a whole branch of HFT that does very complex versions of this.

I know, but these guys also only cook with water. My impression is that their ideas are limited by the tools they are using (statistics and machine learning software). It is in my opinion not inconceivable that outsiders, even single persons, can still find profitable models that wouldn't come to the mind of the pros.

[Copper]

Sure, if you want to beat the other guys, you have to do something different from the other guys.

[89th]

@Copper said in I'm no stock expert, but:

Sure, if you want to beat the other guys, you have to do something different from the other guys.

I don't care about beating the other guys. I only care if my investment makes money.

The market was up today. It will be down tomorrow. I GUARANTEE IT or I'll close up my new investment firm, I promise.

[Horace]

@89th said in I'm no stock expert, but:

@Copper said in I'm no stock expert, but:

Sure, if you want to beat the other guys, you have to do something different from the other guys.

I don't care about beating the other guys. I only care if my investment makes money.

I suggest treasuries. AAA rated bonds are for adrenaline junkies with no regard for their own financial safety.

[89th]

Ok I’ve decided to close my investment firm. The market was up two days in a row.

[Horace]

@89th said in I'm no stock expert, but:

Ok I’ve decided to close my investment firm. The market was up two days in a row.

Treasuries earn money every day, no exceptions.