Very nice. So what kind of a return is that $2.5k for you then? Wasn't sure what kind of buying power you have in your portfolio so I'm interested to hear what kind of profit percentage that is. Thanks!

suit 

Very nice. So what kind of a return is that $2.5k for you then? Wasn't sure what kind of buying power you have in your portfolio so I'm interested to hear what kind of profit percentage that is. Thanks!

To date I've intentionally not addressed capital, but did plan to address it in the future. Since you've asked the question though I will try and address it now.

From iSystems...
REQUIRED CAPITAL. Required Capital is the amount which must be in the account in order to activate the system for live trading, and is based on the contract's margin plus the worst historical session loss for the system.
SUGGESTED CAPITAL. Suggested Capital is a calculated value which uses the system's worse historical drawdown and volatility to target a future drawdown less than -33%.

Note that this numbers are all lower than exchange margin requirements, which is what your broker statement will show.

For this portfolio of 6 systems Required Capital is $27k and Suggested Capital is just under $200k.

Even then account ROI isn't necessarily your actual ROI. I was listening to a presentation recently when the presenter made the point that due to the recent Refco, PFG & MF Global failures he now only keeps 25% of his trading capital in his trading accounts and regularly sweeps from and to a bank account to maintain that percentage. A few minutes later he was claiming that a certain account he was trading was up X% percent. Question is what that based upon the amount in the account (it was) or was it based upon his dedicated trading capital. In reality this presenters claimed returns where X/4% and not X%.

Due to the way intra-day margining works, the Required Capital numbers are very low. The problem with the Suggested Capital numbers is that they are individual system based and not portfolio based.

To illustrate this let me use a simple example of 4 systems, that each make the same amount of money. In this example Sharpe is just Average/Std Dev,



Now let's combine these 4 systems into 6 different portfolio's that each contain 2 systems each.



Not surprisingly all portfolio's make the same amount of money. The portfolio with the best Sharpe ratio is the portfolio that combines systems 2 and 4 and almost goes straight up. That might seem obvious since the two individual systems with the best Sharpe ratio's were 2 and 4. But the 2nd best portfolio is the combination of 1 and 4, and system 1 has the worst Sharpe ratio. The worse portfolio is 1 and 2, which may surprise people given that 2 has one of the best Sharpe's and is also included in the best portfolio.

So why is this?

Correlation.

Positive correlated systems amplify movements, while negative correlated ones mute them. So as long as systems have positive expectations you want to have uncorrelated systems rather than correlated ones. (Which is why I mentioned the correlations of my 6 systems in one of my first posts!)



As you can see the portfolios with the highest Sharpes were the portfolios that combined uncorrelated systems, while the portfolio's with the worst Sharpes combined correlated systems! As such the $200k Suggested Capital mentioned above would only be correct if the 6 systems in my portfolio all had a correlation of 1. Since my portfolio has an average portfolio of 0, this is obviously not a good metric in anyway.

Another way to illustrate this is to look at the 'modeled' monthly drawdowns of my 6 systems, and compare that to the 'modeled' monthly drawdown of my portfolio. The individual systems have a max drawdown of between $2k and $8k each, and sum to over $30k. The max drawdown of my portfolio is less than $5k though, or 14.9% of the sum.

When I talked about Capital, I wanted to also talk about risk, and how instead of calculating an ROI, you can design a portfolio to a specific risk profile. Time is limited though so I will try and do that next time.

NOTE / WARNING :- iSystems in their Suggested Capital definition finish with "No guarantee is made that the system will not lose significantly more than 33% of the suggested capital amount'. The obvious flaw with any of this analysis is that it is looking at historical data and using it to imply performance/risk into the future. Not only is this dangerous for obvious reasons, but this is doubly dangerous in this case due to the limited nature of our historical data. Most of my analysis only goes back to 2010 and as such excludes the 2007/8 financial crises, the 1998 Russian Debt Crisis and implosion of LTCM and the 1987 Black Monday crash to name just a few. As such we have no idea how any of these systems would behave in situations like those, never mind potential disruptions in the future. Just like 'Out of Sample' or 'Since Tracked' average performance was below the backtest, you will probably find that 'Out of Sample' risk is higher.

Saturday 26th August – End of Week 3.

Note:- last week should have read Saturday 19th not 29th. As much as I would like to, I can’t time travel yet.

The week started with an Eclipse and ended in a Hurricane.

Unfortunately for this account, the week started with losses, continued with losses, followed by more losses, with Friday being the only positive mark to market day. All 6 of the systems have now traded. Two are doing great, one well, one is breakeven and the last two are doing nearly as bad as the best two! Made almost another €1k but lost almost $2k, for a net loss of approximately $1k, meaning I gave back about 40% of my profit, and profits through 3 weeks are now approximately $1.5k.

With a monthly mean PnL of ~$4+k* and a standard deviation of ~$4k I would expect weekly PnL to average ~$1k with a standard deviation of <$2k**. So the majority of weeks PnL should be between <$1k> and $3k, so this loss still falls within expectations.

Plan to re-review everything in the next few days, and confirm system selection for next month. This has nothing to do with this week’s losses and will be an end month process every month. Not looking to chop and change every month so I doubt I will be removing any systems. I think I may be adding one or two more systems bringing me up to seven or eight. I know there’s at least one system that looks good that didn’t meet my minimum ‘since tracked‘ threshold last month, but will do this month. It also trades a contract not currently present in my portfolio.

* Since tracked numbers. Full backtest has slightly higher mean and significantly higher standard deviation.
**Assuming a 21 day month, so weekly mean is monthly * 5/21 and weekly Std Dev is monthly / square root (21/5)

SMCJB 

Plan to re-review everything in the next few days, and confirm system selection for next month. This has nothing to do with this week’s losses and will be an end month process every month. Not looking to chop and change every month so I doubt I will be removing any systems.

Of the two systems doing badly, one has had a great week and is now slightly positive, while the other hasn't done a trade. While I wasn't planning on dropping any systems upon further analysis I am going to. My one bad performing system this month experienced a loss
- equal to 1.5 times it's largest drawdown ever
- more than double it's worse month previously
- that is 3 standard deviations away from its average monthly PnL.

Also it turns out that this system is the one with the shortest 'Tracked Record' of the six I'm trading. Coincidence or lesson to learn?

So deactivating one of the systems today before month end. Leaves me with 5 systems right now.

August Recap.

Posting as a picture since tables don't work very well.
Pretty good start.

Sunday 17th September - Mid Month Update

Activated 3 systems at the beginning of the month, to bring the number of systems traded to 8. First two weeks of September don't look good. One of the newly added DAX systems in particular had a nasty -€2500 day trade which pretty much wiped out August's profit's. Since then it's been one step forward and one step back. 6 week trading PnL is now ever so slightly negative but when you add in system licence's I believe I'm down about $1k. While this might seem concerning, everything is still firmly within expectations, so while I'm obviously disappointed about the results, I'm not concerned.

Saturday 23rd September

Unplanned update. I've been travelling for 3 days (actually at @kevinkdog's Strategy Factory Workshop) so I haven't been following the performance of individual systems, but I could see the PnL wasnt doing well though. Now down over $3k (so $6+k this month). Seems like 6 of the 8 systems had bad weeks! . Month end is next Saturday so if I wanted to switch off any systems I will have to do it by Friday.

SMCJB 

Saturday 23rd September Unplanned update. I've been travelling for 3 days (actually at @kevinkdog's Strategy Factory Workshop) so I haven't been following the performance of individual systems, but I could see the PnL wasnt doing well though. Now down over $3k (so $6+k this month). Seems like 6 of the 8 systems had bad weeks! . Month end is next Saturday so if I wanted to switch off any systems I will have to do it by Friday.

FYI to @Big Mike - apparently the popularity of futures.io runs deep, and worldwide. At least 9 of us this weekend (from 4 different countries) in Cleveland are futures.io members!

SMCJB 

Saturday 23rd September Unplanned update. I've been travelling for 3 days (actually at @kevinkdog's Strategy Factory Workshop) so I haven't been following the performance of individual systems, but I could see the PnL wasnt doing well though. Now down over $3k (so $6+k this month). Seems like 6 of the 8 systems had bad weeks! . Month end is next Saturday so if I wanted to switch off any systems I will have to do it by Friday.

I am working on another approach to actually switch on systems first when they are loosing .Since all systems will have loosing perdiods no matter how good they are. I am trying to use consecutive winners and loosers and standard deviation using the equity curve from the strategy to determain when to start trading it. I always launched strategies when they are doing good and then they go in to drawdown almost immediately. But since it is not the strategy that changes it is the market so it depends on the type of strategy and Kevin D got it nailed by using WFO do adapt them which You cant do if You rent a system.

Attached is a quick and dirty Analysis I did of the iSystems programs:



It's easy to do, you just export the data from their site into excel and add it up. Takes minutes and will save you $.

From the Total P&L and Tracked P&L's, you can see that, over time, profit diminishes, a lot.

And from the Live P&L's you get a total return of only 1.7% for letting these boys put your investment at risk for over 3 years with some incredible volatility and draw downs.

You really have to be a wizard to determine which of these systems are working and which are not as it appears that you have to find a system out of the 950+ systems that happen to be working at the moment and over time the chance seems to become less and less.