Faith In Research - Post mortem

the faith in research trader demonstrates a track record of skill.

my opinion is to remove the trade if it will result in significant system disruption due to a mistake.

doing so does not in my opinion reduce transparency and C2 credibility. In fact, it enhances C2 credibility demonstrating the C2 platform’s ability to handle realistic events. As the C2 platform improves simulated trading, one can expect more errors. In fact, ideas that seemed appropriate in the past will become obsolete as innovation moves forward.

i think the C2 decision is valid and the C2 operation should be free to move forward from this issue.

i would like to propose that the system prevent trades relative to margin requirements.

for the C2 platform to govern a developer system trading size is inappropriate without the developer’s permission. doing this can actually increase a system’s execution risk, tactical trading risk, and other variables as well as reduce the system’s simulation validity.

Mathew - It is my opinion in interests of full disclosure, just as most brokers still leave an entry in your transaction / order history, the trade although busted, cancelled or house intervened (in this case) should still be listed in the systems transaction history as such with a comment/popover/info icon or whatever indicating the reason the trade did not happen. You could give the key to the reason(s) in the help. If you omit these things from the history who’s to say what else is ommitted - its a slippery slope.

FWIW - i admire your honesty on coming out and bringing this up - says alot about you !

The purpose of C2 is to provide a realistic simulation of real trading. C2 has gone to great lengths to try and faithfully match real trading execution in their hypothetical fill engine. In this case, real accounts brokerage firms refused the bogus order because they have checks and balances which saw it as a probable mistake. I think C2 did the right thing also deleting the trade, even though they didn’t have the same checks in place. Now they will add the first set of those checks, and probably more over time. The intent of C2 is to realistically match what happens with live brokers; it is an iterative process and I don’t find anything wrong with changing a fill that C2 gave, but now realizes wouldn’t have happened in a live account.

A while ago, Matthew implemented a very intelligent feature (still in beta mode) that allows C2 vendors to voluntary set trading constraints on their own trading systems.

Developers can now choose a value for the following restrictions:

1: No single position may take more than % of system capital

2: Sum of all open positions may take no more than % of system capital

3: Trades must have a stop loss capping loss at this % of system equity (C2 will create one if you don’t)

(This feature can be found in ADMIN—>Constraints)

These simple safeguards could prevent extremely large and/or unusual orders/stops from ever entering the trading platform.

I thought the intent of C2 regarding historical system information was to provide subscribers with history of a system so they could make an informed decicion as to whether or not to subscribe to a systems decisions. As a subscriber i would certainly like to know if system owner had a susceptibility to being accident prone. Remember those trades bogus or not were still sent to the subscribers as and how as a subscriber would you know its bogus ? some poeples accounts are bigger than others.

Remember those trades bogus or not were still sent to the subscribers as and how as a subscriber would you know its bogus ? Some people accounts are bigger than others.

Seriously, you think there is a C2 trader rich (and crazy) enough to put over 12 MILLION dollars margin money just to buy 2419 mini S&P 500 futures contracts because a C2 vendor said so? :slight_smile:

You dont have $12mil ? - your behind pal :slight_smile: - the point i’m trying to make is that it could have been 24 contracts or 241 or 29 anyone of which could have been in error. The magnitude of the size is unimportant its the fact that it happened, it was a live trade, the system informed subscribers of the signal and they could have taken the trade. It should be logged in the transaction history.

I bet there’s many vendors out here now who would like to call mathew about an erroneous large bet they made that killed their system that had no real trades …

"You dont have $12mil ? - your behind pal :)"

Well I just lended 3 billion dollars to Soros so I only have 11 millions left. :slight_smile:

Joke aside, I think Matthew will probably come up with some cool feature that will prevent this kind of trading error from ever happening again, even though nobody was "hurt" in this little incident.

For example a small window would pop up and say : "Whoa, easy with the Vodka cowboy, are you REALLY sure you want to short 3,000 contracts of Soybeans?" :slight_smile:

Igor, how do you know that no one was hurt? What about someone trading manually that Matthew knows nothing about.

Where do you draw the line at erasing bad trades? Is it ok to erase 12 million but…

1 million? 100K, 10K, 100

“Igor, how do you know that no one was hurt? What about someone trading manually that Matthew knows nothing about.”

Well, nobody complained so far. For the auto traders this weird C2 order was immediately rejected by the brokers anyway, so no harm was done. The manual traders ? I guess they simply laughed at the ridiculous size of the position and ignored the signal.

Come on guys, you are blowing this freak and rare incident way out of proportion, let’s simply learn from it and move on.

"Whoa, easy with the Vodka cowboy, are you REALLY sure you want to short 3,000 contracts of Soybeans?"

Thanks for the laugh ! - that was funny !

The first review about this system says it all: ” a lot of mixed results. A gain in one week can be virtually wiped out in one day.” Then another review: “Can’t seem to figure out whether it’s coming or going.” There was the 55% drawdown in the first month of its existence, and what looks like another 50-55% drawdown just a month ago. This trade “typo” a couple days ago is just another fact about this vendor, consistent with the other factual evidence, and the drawdown should be recorded just like every other statistic to protect anybody from ever putting real money in this system. Apparently, the two 55% drawdowns didn’t drive innocent people away, so maybe the 130% drawdown will.

I agree that it’s unfair to tell Kevin Davey and other vendors who have made mistakes “tough luck,” then for some arbitrary reason tell the vendor of Faith in Research we’ll roll back your mistake.

Anyone who calls his system Faith in Research should agree with me and others: let us have all the facts so we can have FAITH in our ability to properly RESEARCH a system.

Some further details on this, after more research, and some further thoughts:

The signal which was mistyped by the vendor and was a huge quantity was a CLOSING order. The instruction emailed by C2 was “Sell To Close 2417 contracts” (for those of you unfamiliar with that lingo, it means Sell 2417 contracts to close your currently open long position). The position size at the time was 2. The idea that some manual (non-AutoTrader) trader would see the email and then meekly type in an order to Sell 2000 contracts, just because an email had the words “Sell 2000” strikes me as silly.

Now, on to what actually happened in real life. No AutoTrader traded this because it was technically impossible to close the position of 2417 contracts, because no one had a position of 2417 contracts to close.

(In other words, the real error here was C2’s accepting a closing order for a position that didn’t exist. This has since been changed; more on this in a moment.)

As far as checking for overuse of margin: This is why we didn’t reject the order as using too much margin – because it was an order to close a position. We don’t check whether an order will fit within margin constraints when it is a closing order, since closing orders reduce margin use. (Had it been an order to open a position, the order would have been rejected as requiring too much margin, of course.)

Regardless of all that, had anyone actually traded this, in any account, due either to a software error or due to human stupidity, we most certainly would be obligated to include the trade in the record.

Had the order been a “Buy To Open a new position” of 2417 contracts, then we most certainly would have been obligated to include the trade in the record – no question about it.

But no one did trade it, and the trade that was emailed to people (“Buy To Close 2417 contracts”) was clearly an error. There was no position to close.

I do feel embarrassed that C2 even allowed such an order to be typed in, and that C2 emailed the order to people, and that C2’s hypothetical track record for a brief time said the order had been filled by the C2 Fill Engine. But it wasn’t filled anywhere, and really couldn’t have been.

In response to this incident, we’ve implemented some new “fat-thumbs” safe-guards for order entry.

First, if you enter an order that seems more than 2x larger (the ratio varies depending on instrument) than what we typically see in your system, we will issue a warning and ask “Do you really want to do this?” On top of this, we simply flat-out prohibit order sizes above a certain amount, and won’t let you enter such large orders, even if you supposedly “want” to.

Finally, we have implemented a closing-order math check on all closing orders, which is designed to make sure the position quantity actually exists in your system before we even accept the closing order.

So, in closing, here is what I am inclined to do: Let’s treat this as a busted trade (never was fillable in theory, never was filled in fact), and let’s leave this forum thread as a testament to the fat thumbs of the system vendor… and a reminder to traders everywhere that trading, like everything in life, is dangerous and subject to human error and incompetence. (In this case the incompetence was mostly mine, in writing software that allowed the order to be accepted.)


The margin thing isn’t so simple. Some brokers charge as little as $500 margin for daytrading. Take US, the t-bond, that’s currently trading at about 123, has an overnight margin of $3713 and a daytrading margin of $500. If someone with a $70K account entered an order to buy/sell 123 contracts of US at such a broker, would the order get filled? Almost certainly if it were a market order. A stop or limit order, maybe, maybe not.

Matthew, I am glad to see you have made some changes to the system that prevents selling to close a position that you don’t have. I had a very small trade like that once that I wondered why the system allowed me to enter, so I am glad this hole has been filled.


What is the final outcome of this issue. Will the trade stand or be erased?

Based on the fact that the trade in question was marked as an instruction “to close” a position which did not actually exist, the error was C2’s. We allowed the typo to be sent via email (this has now been fixed and won’t be allowed again).

Further, no one traded it.

We will therefore leave it as a busted trade, but this forum thread will stand as testament to the fact that the vendor has fat thumbs, and manually types in trades, and can (on occasion) make mistakes.

Let us be glad that no one traded the instruction, and that we extracted at least some good from the incident: it was a wake-up call to impose sanity-check size constraints on C2 manual order entry.

Sometimes 2 wrongs (entering an order as "sell to close" when no open position existed, switching prices and quantities) actually do make a right!

Knowing these additional facts, I think Matthew made the right decision.

Yes, I agree. Knowing these additional facts… "sell to close" when no open position existed, makes the decision correct IMHO.

Thanks for allowing the discussion here Matthew.