Early beta test: The Grid

I’ve released “The Grid.” It’s a new feature that allows you to slice-and-dice, sort, and filter trading systems in a dynamic and (hopefully) easy way. It’s in very early beta test, so please don’t have a hissy fit if it does something that displeases you. But do give it a try and see what you think.

Click “The Grid” on the left-side menu bar (under Hot Lists).

High Five! Very Nice!

Cool to be able to sort via ADP and thanks also for

Profit Factor.

Really, very nice work.

Thanks! It looks fine. I’m especially happy that we can now sort on RF, MDD, APD etc. Much better than the “simple search”.

If I sort on APD and ignore the very young systems, test systems and systems that have stopped trading, then my impression is that the higher APD region is dominated by stock systems. Do others agree, or do I see what I want to see?

Will this eventually be integrated with the advanced search function, or will it remain a separate functionality? I ask this because it is often desirable to sort somehow at multiple dimensions. I can also imagine other ways to do something like that…for example, where the user can move sliders to indicate the weight that should be given to each dimension in the sorting. Perhaps, if you’re interested, I can work it out exactly some time.

Nice job!

Here are some suggestions:

- explain the meaning of systems appearing in red instead of green

- allow users to specify whether only active systems are shown, or all systems are shown

- allow users to specify that all systems are listed on one page instead of several pages

- allow users to specify some of the advanced statistics to be included (so I can sort all systems based on the Sharpe ratio’s lower bound, or expected shortfall)

-- Yes, Profit Factor is up there.

-- While looking at the grid, I noticed the flying pink pig has coughed up about 6 million in a couple weeks. Is this a C2 record?

MK - at least you didn't call it "The Matrix" (or Reloaded)

It’s interesting to plot the APD against the Sharpe Ratio. I briefly looked at stock systems with more than half a year’s history and a minimum of 30 trades. Excluding a few systems, with a (relatively) high Sharpe (>2) the correlation is pretty high (~0.8).

Looks good except that I can’t find my system in it. Maybe at less than 2 months it is too young?

Stocks possibly, but when I look at all systems & sort by Sharpe (highest first) and looked at the first 3 pages, I don’t really see a strong correlation - loose at best

Taking ALL systems >365 days and >30 trades, the plot still looks interesting. It suggests that the correlation is still substantial (~0.7) for systems with APD < 0.5 and Sharpe <2 (i.e. ~90% of the systems).

The other 10% have an APD > 0.5 but Sharpe < 2 (8 systems), or an APD < 0.5, but Sharpe > 2.

It seems the evidence so far suggests that if you want to have an APD in the longer run > 0.5, you have to accept a Sharpe < 2, and vice versa. We have to see if newer systems (some of which have both APD > 0.5 and Sharpe >2) can maintain that performance as they mature.

Hmm - I would suspect that a number of indicators might align, given enough time on the ‘puree/frappe’ setting. Generally, better systems should probably exhibit more favorable statistics. So, something with a better Sharpe should probably have a higher Profit Factor, for example…

The Grid is useful, but seems fragile. Every time I clicked into a system, the sorted order disappeared when returning.

Interesting. The high correlation is of course to some extent no surprise, since both Sharpe and APD have profit in their numerator and risk in their denominator. But I would not have guessed that it is this high.

The suggestion that a high Sharpe and a high APD don’t go together is interesting too. Why would that be, if it is not just coincidence? Since Sharpe is based on the total equity and APD on the trades, I think that it suggests that focussing too much on intratrade drawdowns has an adverse effect on the total equity drawdowns. (Only for APD > 0.5 of course) But it would be better to compare APD with the Calmar ratio.