Recent content by ast4

Do you have a list of the constituents on a daily basis?

Why don't you backtest against 10 years? Also stress test against major historical events (e.g. lehman, 9/11). As a side note, win rate doesn't mean much without gain/loss per trade.

Oh another thought that just came to mind, the original YQL table I sent you requires that you enter expiration. While it's always 3rd friday, sometimes this can be a pain to generate. Luckily there's another table within YQL that allows you to retrieve all current available contracts. YQL Console

Nancy, Though you'll have to keep in mind that you'll need something a little more robust than excel to really take advantage of this. Personally, I've used a python script w/ open tables that continually appends to a csv file. You can then refactor the data pretty easily using the pandas...

YQL Console There's an example, just query throughout the day and store the data - then convert to OHLC bars of you flavor. If that's too much work then just write a black-scholes UDF in excel, back out the implied forward so you get correct IVs. Then take the high and low of the underlying...

There are a few free-ish options that come to mind: 1) Best "free" source I can think of is scraping yahoo intraday, which can either be done directly off the website or using YQL. Then just convert your data to OHLC bars at your desired frequency. 2) If you know the high and low for the...

Definitely if one could afford it SSD would be nice, especially if many of your problems require random access. Often time though sequential access is what is needed for backtesting type problems and speed benefits there are not worth the price differential. We also run tesla c2075s which...

The most reasonable solution I've come across for this is to actually purchase options data. The granularity of the data will obviously depend on your trading frequency, so it might be expensive. One issue about data though (and this is with anyone), is that none of them actually compute the...

It's not so much processing power that you'll need but rather a beefy PSU. If you want to do it "right". Then have 1 machine dedicated to backtesting and 1 machine dedicated to trading. The actual trading machine doesn't need to at all powerful, but the backtesting machine should depending on...

You're not going to find any decent strategies posted in the public domain for obvious reasons. Two good books on the subject are Inside the Black Box by Narang and Algorithmic Trading and DMA by Johnson. You'll have to be careful with the Narang book and be choosy about what you take in, the...

I don't think there's much point, typically software developers have no real statistical backgrounds and also when new measures come up it's not coded in and you have to revert back to what you were previously using. The closest thing to what you're talking about that I know of is probably...

As for as flexibility goes, I think what would be nice to see is being FIX compliant to achieve broker neutrality and connectors for at the broker level for a strong subset of retailish brokers. Same with plugins for datafeed, as well as a built-in CSV and HDF5 support for data reads (I know...

I have no idea why some retail level traders are preoccupied with having a high spec system. The last two firms I've worked for have used pretty low-end machines for actual trading, at least compared to these ridiculous falcon computers. You're not playing games on this thing. A system with 4...

I've been trying to find a reasonable front end software solution all weekend and have still turned up empty handed unless I want to use bloomberg's EMS or something in that ballpark. So I'll tell you what it's lacking, a complete software suite that does the following: -Broker-agnostic (FIX...