I would like to invite you to comment my follow remark.
"Realism Factor" here on C2 have no sense because it is mostly based on the software (TradeBullet) ability.
I will try to explain my thought. Look at example below:
Realism Factor for the system Pippo:
Volume Factor: 100%
Limit and Stop Factor: 100%
Auto Trade factor: 0%
Avg Slippage: $.0
As % of P/L: 0%
Overall Realism Factor: 0%
As you can see the very major value (slippage, limit and stop etc…) are 100% but the Realism Factor total value is 0% because it is based on the possibility that tradebullet works or do not works. IMHO it have no sense.
What do you think?
No Comments from system developer and/or system follower ?
"As you can see the very major value (slippage, limit and stop etc…) are 100% but the Realism Factor total value is 0% because it is based on the possibility that tradebullet works or do not works. IMHO it have no sense."
It doesnt have anything to do with possibilities. The auto trade factor is based on the actual results of subscribers using TB with a system. If it had a 0% that means that zero out of all of the people using TB with the system got filled on that signal.
I believe the various individual percentages are multiplied together to determine the RF % so that is why a single 0% can cause the total RF% to become zero. MK can confirm or deny that is true.
It would be nice if you could post the exact details of the trade in question so we can see how realistic or not it was. I can’t tell if you are implying that a technical problem caused TB to not get a fill on an otherwise perfectly tradable signal - if that were the case then I agree that this RF factor is unfair but I can’t tell based on the info you provided.
I agree with Mario.
The last two trades in my SnapBack system also received a realism of 0% due to Autotrading. One contract of Notes and two of Bonds, all stop orders and not limit, well within the daily range and both have volume of several 100K, should not get a realism of 0%.
If no one got filled with TB in these trades, then there is clearly something wrong with the C2/TB interface. Systems should not get penalized if C2/TB is not working. System should get penalized if unrealistic orders are displayed as filled.
- Fanus
I think all systems are in this comic situation. System Realism Factor is in fact based on TradeBullet realism factor 
If TradeBullet works you have a great rf else your real trade signals are not "realistic.". It is not important that your signal have right volume, slippage etc… or your limit order was crossed, it is not realistic because TB is not working.
I also know that many of C2 system are autotraded outside C2 too but no one have a 0% autotrade factor because Strategy Runner or TradeStation etc… works very well.
Now. I agree with anyone want to use TB but the Realism Factor should be named "TradeBullet Factor"
The AutoTrading Factor is not determined solely by execution data from TradeBullet. Any C2-Compatible AutoTrading Software (Strategy Runner, Strategy Manager, FX-Auto, etc.) will be included in this statistic.
I do not agree that we should denigrate TradeBullet and claim that it is due to TradeBullet that an order you entered was not executed by your customers. The fact that an order was not filled in real life can be caused by all sorts of reasons. Yes, maybe the end-user failed to turn on TradeBullet. But maybe also you entered the order at a very inconvenient time of day. Or maybe you suggested a limit price that was “just touched” on light volume and not filled in the real-life customer account. (Keep in mind these same events will lower the RF score for the trade, even if the end-user is trading with Fx-Auto or Strategy Manager or some other future C2-compatible product.)
Let’s talk about the why this part of the RF Factor exists. The idea behind this aspect of the Realism Factor (RF) statistic is that, on average (and over a long-enough period of time) the statistic will accurately reflect the “tradability” of a system. In other words, if you consistently enter orders into C2, but those orders consistently are not filled in real-life AutoTrading accounts, then that probably tells a prospective trader something about a system.
I would suggest that focusing on just a handful of individual per-trade RF scores is not really useful. The more important number is the overall system RF number. And that is calculated based on a weighted average of all individual-trade RFs (weighted by trade P/L result).
I am happy to revisit the way RF is calculated, and in particular, to look at the weighting of the AutoTrade Factor within the RF calculation, but I don’t agree with you that the score is useless, that TradeBullet flaws are responsible for a bad score or two, or that there are serious methodological flaws with the current calculations.
First of all, I do not want absolutely denigre TradeBullet.
You have to explain us why this example trade have "Overall Realism Factor = 0"
Volume Factor: 100%
Limit and Stop Factor: 100%
Auto Trade factor: 0%
Avg Slippage: $.0
As % of P/L: 0%
All value (less AutoTrade Factor) are excellent but Overall Realism Factor are 0
Why?
The second thing: No one autotrade software I know must run to receive signals because it are on the server.
I do not know Strategy Runner are C2 compatible, how can I use it?
Matthew
Based on the last couple of trades in my Treasury Trader and Snapback system, Tradebullet/Autotrading effectively WAS the only factor determining the RF value.
All other metrics was 100%, except Autotrade which was 0%. As a result, the RF is 0%. The orders was for at most two contracts in the bonds and one contract for notes, which is extremely liquid, the orders was during regular hours, which is definitely not an inconvenient time. And frankly, if someone use Tradebullte for autotrading, then there shouldn’t be an inconvenient time. Isn’t the whole purpose of autotrading to place orders when not convenient for a subscriber?
Anyone placing the trades receiving 0% RF in my systems, would have been filled within a tick, or two if not at the fill price. The only reason for it not to be filled was that Tradebullet was turned off, or did not work.
I agree with Mario, that RF is 100% based on Tradebullet and if Tradebullet for whatever reason, doesn’t work, the system get penalized, no matter how realistic the trade is.
If Tradebullet reports are included in reports, only attempted trades by Tradebullet should be included. Only after Tradebullet informed C2 that it attempted a trade and provide some kind of reason why it wasn’t filled, should it be included in RF.
Regards
- Fanus
Fanus, you have to believe me I check many system on C2 and ALL are in the same situation.
Trades absolutely realistc get "0", lowering all the rf average.
The sad thing is that many C2 customers believe in that average and they ask me how can trade the system in different way or they do not want to trade it because RF is less than 100%.
"Yes, maybe the end-user failed to turn on TradeBullet"
MK, was this a mis-statement on your part or did I mis-interterpret what you are saying? I’m assuming that the autotrade factor is based on users who are actually using TB (or other autotrade software) - a user who has TB turned off (or otherwise misconfigured to prevent a trade from executing) would not be factored into the autotrade rating - true?
Excellent point guys. I thought I was the only one struggling with issues regarding RF. I have presented the developer on several occasions solid proof why the RF needs to be restructured to properly reflect real life fills, yet to this date nothing has been done about it. This issue has been discussed over and over for the last 6 months. Just look at my closed log and you’ll see. The best way is to put a statement in your system description. Intelligent potential subscribers will understand and appreciate your point of view. The ones who are serious and experienced will join. Also vendors should have the right to decline not having an illogical value be displayed on top of their stats and instead that value be replaced by a text saying something like “Vendor declines RF value being presented, see explanation”. Shouldn’t the RF stand for “Ridiculous Formulas” until it reflects what really takes place in the real world of trading?
Regards,
Tarek
How is the autotrade factor defined? Is it defined as the percentage autotraders that get filled? Then, if no one turns tradebullet on, it should be n/a instead of 0%. Then it would be impossible to have an autotrade factor greater than 100%, but extreme-os has a trade with autotrade factor 102%.
Is it defined as the average percentage profit that autotraders have, relative to the C2 outcomes? Then it is strange that the outcome is 0% so often. I mean, 10.32% or -8.94% is something I would believe, but 0.00%?
How do the trading permissions influence the autotrade factor? If a subscriber does not give permission to trade options, would that decrease the realism factor for option trades of the system? That wouldn’t be fair.
How does the broker influence the autotrade factor? I have experience with a system that sent many signals to short stocks, but in most cases these stocks were not shortable at my broker (MBT). They were shortable at another broker, but there autotrading was not supported. That means that autotraders effectively have no profit from such trades. Similarly, if all autotraders of a system use IB, and certain futures or options are not supported by IB, then it can be defended that these trades get an autotrade factor of 0% (unless it was clearly communicated that one is supposed to use another broker).
Jules
The broker issue is probably not the cause in the case of WhyNot, because both 0% and 100% happen with @USM6.
Tarek
I’m sorry to say it but it seems that Matthew concentrates its attention too much on the customers request, omitting completely the system developers request and comments.
Every time I have asked an improvement (also considering it as a good thing) has been said me “I will make it very soon” but actually it never been made.
I think we are part of this collectivity too with our suggestions to improve together C2.
Am I wronging ?
Yes Mario, I understand how you feel. Here is the most recent example from my trade log (closed a few moments ago). I just closed a spread trade on TIE (Titanium Metal), one of the most active stocks on the New York stock exchange over the last week. Those options (June 30 Calls and June 35 Calls) already traded over 4500 just today on each strike and an average of over 53,000 according to IVolatility.com. Yet they received a mere 7% and 18% of realism respectively. Hmmmmmmmmm.
Tarek: That’s disingenuous. The trade you posted which you complain about was:
BUY 100 TIEFF at the market
Now, the TIEFF call is traded heavily for an option, but market makers are typically posting bid/asks of quantity 30 or so. Thus a market order to buy 100 would not get filled entirely at the C2 price. In real life, only 30 would be filled at the fill price your received. Perhaps another 30 (maybe less) would be filled at a nickel higher. And then another 30 (maybe less) would be filled at another nickel higher.
So, we’re talking about a real-life fill that will almost certainly be higher than the fill you received on C2.
Now, let’s assume you had five or six subscribers. Suddenly, a market order to buy 100 would turn into a market order to buy 500. And while its true that many (most?) subscribers would not trade with $100,000 accounts, at least some might.
Now, I’m not saying that the RF is perfect, but it does at least indicate to potential subscribers that their fills will not be exactly the same as C2’s hypothetical fills. It’s not designed as a punishment to you or other system developers. It’s just a math formula designed to help analyze systems and their real-life tradeability.
Is 17% RF for that particular trade too harsh? Maybe. But it’s based on a math formula that seems sensible to me. That’s all. Let’s not invest it with some deeply significant meaning or importance that it does not have.
And - finally - this last message is not directed particularly at you - but is rather intended as a more general point. If you are a system vendor and you think the RF factor for a trade you make is totally inaccurate, the fastest way to convince me of that fact is to show me a real-life brokerage statement with real fills at the same exact price and quantity as the hypothetical fills on C2. Then I’ll know that the low RF number I have assigned is totally inaccurate. But until then, the best I can say is: "I agree the RF algorithm is not perfect, but it’s a good starting point."
Remember, I’m perfectly happy to take vendor feedback into account, and I will continue evolving the RF over time. The RF is not set in stone. And it’s not designed as a mean-spirited punishment to keep system vendors unhappy.
…It’s not designed as a punishment to you or other system developers. It’s just a math formula designed to help analyze systems and their real-life tradeability…
Unfortunately potential subscribers are actually scared from a system with low RF.
I received many emails from potential subs ask me how is possible that trade received “Overall Realism Factor = 0” if the limit order was crossed, the TBond is very liquid, there is no slippage etc…
How I explain it?
Now, I’m not saying that the RF is perfect, but it does at least indicate to potential subscribers that their fills will not be exactly the same as C2’s hypothetical fills.
It is not true because potential subs perceive a low RF as an unrealistic system and not "that their fills will not be exactly the same as C2’s hypothetical fills."
But it’s based on a math formula that seems sensible to me. That’s all. Let’s not invest it with some deeply significant meaning or importance that it does not have.
Unfortunately (believe me) potential subs do it, it is essential for most of them.
Ok I’m back. You said “Now, the TIEFF call is traded heavily for an option, but market makers are typically posting bid/asks of quantity 30 or so.” Please Matthew, with all due respect man, I may have been born at night but I wasn’t born last night. Do you have proof that that is the norm and not the exception? Could you please show us the proof that MM’s are showing 30 or so the majority of the time. You may have seen that advertised for a brief moment when I reported the issue on the forum but that is not the norm on a high velocity stock like TIE. The sizes available are rapidly changing on each side as these MM’s have to compete with each other to make money for their firm. I tell you what, how about if I do this to give you more proof than I already have in the last 6 months. How about if I take snap shots of those June options tomorrow and I’ll even let you pick the time of the day, then I will display the screen shot for everyone to see. In fact, people can do it on their own and load up the option chain with any broker and see for themselves the average available size on them at various times of the day.
Also, asking for brokerage statements doesn’t proof or convince anyone what the available liquidity was the exact moment when the order was executed and time stamped. Can you show me a brokerage statement that does that?? I’m sorry but I’m not quite following your logic here.
In addition, how do you justify the other trades that received a 100% realism factor (GCI options, 600 babies, all at the market) when it should be lower, but another received 11% (UNH options, 75 of them, again at the market) a very active issue and options that is part of the S&P 400 and I think the S&P 500??
Please understand that my arguments about this critical issue are in no way an attempt to pick on you in front of everyone or god forbid, belittle your intelligence. On the contrary Matthew, I like to help out with actionable ideas that would make this site trader friendly and ultimately benefit everyone. After all, my interests are the same as yours. Is that so wrong? You see, this issue has been on the back burner for a good 6 months and you said to me on several occasions that you will be improving it. But.
.
Well, I wish you the very best.
Regards,
Tarek
"Now, let's assume you had five or six subscribers. Suddenly, a market order to buy 100 would turn into a market order to buy 500."
I'm with you. For example a 200 lot in the DJM could take over an hour to fill. Assuming five subscribers that "market order" could take several hours to fill. Moreover, I doubt anyone is auto trading signals that size off C2. So that system ("Test System"), and others doing similar size, get a free ride in regards to "actual" fills. Now WhyNot is trading relatively small size (less than 20 contracts) on the @US which has volume 100's of times or more the the DJ. he is using stops and limits. The stops are effectively market orders after their price is hit. Limits that are penetrated should have 100% realism. However, these systems have relatively similar RF's. I'm not trading these, but I have traded these markets, and I don't think the RF is real in this example.
Test System Realism Factor 76.6
STO 200 DJM6 11435 4/27/06 12:05
WhyNot Realism Factor 82.0
BTO 17 @USM6 107 1/32 5/23/06 10:00
