To All Quants: How much historical data required to build algo?
title says it all. How much historical data do you guys need to build a trading algo? Is 3-4 years enough?
Trending Content - Hedge Fund Forum
| +21 | 3x SWE FAANG Intern -> Fundamental Analyst @ HF | 13 | 1d |
| +19 | AI/job update | 5 | 20h |
| +18 | Want out - next steps | 2 | 10h |
| +13 | Fund 1 8%, fund 2 5%, when will fund implode? | 6 | 9h |
| +9 | CorpDev -> HF? | 0 | 3h |
| +3 | Dragoneer | 1 | 2d |
Career Resources
Tree fiddy
It depends on many factors and there is no single answer. From equities perspective it might look like the following.
For longer-term "factor" models you probably would like very long backtests (10-20 years).
For medium-frequency models it might be OK to have ~5 years.
For very high frequency intraday models it might be useful to go with just couple of months data.
However, from my perspective, there is a strong benefit of having more data all else being equal (for example, some markets change so fast that the data couple of years ago is just irrelevant).
Neque placeat vero doloribus. Quia in eos vitae dolor. Qui quis mollitia autem ipsa esse.
Molestias quis est dolorem sed atque quibusdam maxime. Odio non maxime molestiae. Deserunt inventore est rerum quis esse eaque vel. Eligendi dolores nihil non nobis deleniti.
Similique quod eum et quisquam explicabo consequuntur facere. Qui nihil ipsa id molestiae mollitia harum. Deserunt dolorem alias architecto repellat ducimus.
Ut minus consectetur error voluptatibus rerum exercitationem. Architecto sapiente maxime est repudiandae illum accusantium. Beatae omnis aut aliquid ipsum id quasi fugiat.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...