Choosing Tools for Trading Research, Backtesting, and Execution
Summary
The document surveys choices for researching, backtesting, and running trading algorithms, including custom code and dedicated platforms. One contributor favors building a research workflow around personally managed data and scripts, arguing that this makes test assumptions easier to inspect. Other responses mention R, Python, Java, Matlab, and platform-based tools, with preferences shaped by programming flexibility, data size, stability, and execution needs.
A recurring lesson is that historical data, backtest logic, live data, and execution code are connected, so researchers should understand how each component affects results. Practical considerations include data quality, adjusted and delisted security histories, storage formats, processing speed, charting, and simulation output. The examples are individual experiences rather than controlled comparisons, and platform capabilities and costs may change. The discussion also distinguishes daily-bar research from intraday work, so no single tool choice is presented as suitable for every strategy or scale.
Key ideas
- Custom research code can make backtest assumptions easier to inspect and understand.
- Historical data, simulation logic, live feeds, and execution systems should be considered together.
- Data quality includes adjusted prices and records for securities that no longer trade.
- Language and storage choices depend on data volume, speed, and research needs.
- The platform recommendations reflect personal experience and may not generalize across trading styles.
Tags
Full text
# Delta of a Commodity Future
# Delta of a Commodity Future
Generally the price of a future is
$ F(t,T) = S(t)e^{r(T-t)}, $
and it's delta is:
$ \frac{\partial F}{\partial S} = e^{r(T-t)}. $
(As opposed to the delta of a forward which is always one.)
In some explanations this is proven by taking the derivative of the pricing equation above see RRGs response in this link:
Are Futures exactly Delta One?
However for commodity futures the pricing can incorporate a convenience yield (c) that can give rise to backwardation.
$ F(t,T) = S(t)e^{(r+c)(T-t)}, $
However what should the delta of a commodity future be?
If we take the derivative of the pricing equation it would be:
$ \frac{\partial F}{\partial S} = e^{(r+c)(T-t)}. $
Or....
Since the difference in the delta of the futures and forward prices are purely down to the way that futures are settled each day (and nothing else), and so the delta should be that shown below, the same as any other future.
$ \frac{\partial F}{\partial S} = e^{(r)(T-t)}. $
Q1.) Is it correct that the delta of a commodity future is the same as any other future?
Q2.) Is this the delta I should use when hedging? i.e. hedging 1 1m futures contract with 1 12m contract will not be completely delta neutral instead I should apply the discount shown above to get the correct hedge?
## Answer by Alex Bădoi (score 2)
https://quant.stackexchange.com/a/22956
Since all futures are linear instruments you can achieve a perfect hedge by going short or long into the same future depending on your position.
If however there are no available futures you can use cross-hedging as explained by Hull (2007)
i get an error bellow I'm not sure why so I'll put it in code format:
```
> To answer your question the delta of a future is not perfectly 1
> because in order to for example hedge $10 exposure today for 2 periods
> you would pay $10*e^(0.02*2) for the future. Assuming 0.02r. so the
> delta is 10/10.41
```Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.