Hello, I am putting together a quant strategy for a 1-week hold/trade and am thinking of adding historical RSI values to narrow my results. I was wondering if anyone had any suggestions on the parameters to use in this case? Perhaps, would it be wise to use the 5 period (daily bars) to match the holding time? I am asking for RSI because I know I can acquire those values for 10 years +. If anyone has a better indicator recommendation for a 1-week trade, I would love to hear. 1-week meaning 5-day or 4 with a holiday. Thanks for all everyone.
Since you're a "quant strategist," can't you test various RSI values yourself? The computation of the indicator is very simple.
Calculating this yourself is the way to go if you want to analyze such strategies. Backtesting won't necessarily solve everything, but can weed out obvious loser logic and even non-obvious bugs. 10 years of data is a bit short for EOD data. The longer the better, and not too hard to get some data from around Y2K onwards.
If you want to have a holding period of 5 days then the easiest solution would be to record the time that you open the position and release the position for closure (if so desired) 5*24 hours later. Thus a sort of lock on the position. Isn't that much easier than using an RSI or MACD or other indicator?
Need a lot more info on where you're aiming... Is this for entrance? Exit? Long-only? Long-or-short? Is this a highly-variable market? (NDX-ish) or more staid (DJ30)? Is this a single entity? Or a pairs trade? I have done what you describe for years with SPX options; thinking about it for a single contract is interesting. FWIW, I have never found value in the RSI, but favor Lang's Stochastic, Quong/Soudak Money Flow Index (basically a volume-weighted Stochastic), and Wilder's ADX/DMIs. While I favor the Stochastic/MFI redundancy for intra-day charts, the ADX/DMIs do well regardless of timeframe. As far as look-back parameters go (your actual question! ), you would want the parameters that best match the data in front of you, regardless of your anticipated holding period. Your use of an indicator is to clarify the market before you, and thereby best inform your opinion going forward. (I mean, "Right?") Lastly, never take any indicator's as-arrived parameters as at all fixed, optimised, immutable, magical, or anything else. Tweak them to get the most insight into the market before you, then re-visit the question every 6 months. (Or, just send me your account balance. I have set up a 501.C.3 for that purpose.....)
There is no "best" as the swing ranges change, so do the indicators as they can't keep up as they lag the price, so it is best to redesign RSI or whatever indicator to keep some components and yet add in acceleration and volume formulas. While speed is required for sustained trends, too much speed and lack of volume often sets up for reversals or "retracements". I use a few indicators for scalping ES manually with the above, often times gets in early but this is as planned, and parameters have changed as much as twice a day. Automation does an hourly back tests and readjusts parameters to be weighted more on recent data. And for stocks and daily/weekly timeframes, If there is data for past 50 years, back tests are done or whatever max is on stocks. Each have there own personalities. Holding of a week or less, rather just do weekly options buying ITM.
Is TA and quant trading really mutual exclusive? I thought quant was basically just crunching back price action, etc. to try and come up with strategies that have been profitable over time? I would have thought RSI, and other technical indicators, can fall in the quant universe that they might backtest for. In fact I've heard several "quant traders" on "Better Systems Trader" talk about RSI specifically (not that that is conclusive). Just curious. Thanks!
No; not necessarily. No. This is correct. "Quants" in institutional trading (though it's arguable, especially these days, that independent traders can also be "quants" in their approach) are typically researching potential tricks for HFT's, algos and arbitrage opportunities. They use more technical jargon than I do, but I don't mind them as long as they don't drink my wine.