We made the following changes for tonight’s reconciliation of R2G recommendations with today’s closing prices:
1 - The price for the transaction will be today’s (Hi + Lo + 2*Close)/4 +or- slippage. The slippage is calculated as a per share amount using the variable slippage algorithm, and it’s either added to the price for buys&covers, or subtracted form the price for sells&shorts. There’s been some back and forth on this so let us know actual results if this is not a good approximation. We chose to incorporate everything in the price (commission, slippage) since this is the only thing an R2G sub sees.
2 - A recommendation will NOT be “executed” if the stock did not have volume today. This could be because of three reasons:
a) Stock is suspended
b) Stock did not trade as it’s very illiquid
c) Stock is not longer active due to merger
The above will only necessitate our intervention in the case of mergers (process t.b.d. ) . We will monitor recommendations that are thrown out to make sure all cases are handled properly and realistically, although we don’t expect anything out of the ordinary tonight.
Marco,
The buy price of all the three positions on Monday for my R2G TWY 10 Stocks $20m to $5B were extremely high (8.6% to 11% higher than the highest price of each stock on monday) … I think there is something wrong with this. Could you help to investigate ?
I applaud the efforts at more ‘conservative’ and realistic ‘fill price’ options. The autotrading ‘actual fills’ will, over time, give us a much better sense of slippage and ‘best simulation settings’.
The greatest impact of these new, more conservative settings (in most cases), will be seen in the highest turn, lower liquidity (or higher intra day vol) sim’s. A few notes:
P123 users would ideally have the ability to backtest with any new ‘fill’ formula (although, right now it is likely just average of running 2 sim’s - 1 at Avg Hi//low and one at next close).
More ‘fill price’ options for backtesting would be great. If we had the option to backtest with buys at next day hi and sells at next day low, we could get a better sense on our own of sensitivity of the sim. Steve’s mentioned this several times, and I’d love to run tests with this. I am not advocating using this for R2G’s. I would just like to have it.
VWAP would be nice addition. I assume it’s not doable? Oliver’s posted on this many times.
These new settings will likely give better fills on buys than actual in downtrending markets and on downtrending stocks (as open is not being used). Overall, they may still be fine, but they will have this general bias for people who trade limit orders at open.
Wuu , you are correct. There were still some R2G’s that were using the old fixed slippage percent. I forgot to set all stock R2G’s to variable slippage before launching the new algorithm. I’ll recalculate prices and redo statistics. Thanks
Error in entry prices for Monday in naster 100 . Take FDO a sp500 stock bought at 69.7 market 58/59 . Something went wrong with all the buys on this portfolio .
Since P123 won’t know “if the stock did not have volume today” until the Monday night update, and the email trade recommendations come out before the open on Monday, the subs won’t know you are going to intervene, and the subs will try to buy stocks that would later “NOT be executed”. Are you only going to intervene to not make the trade in the Port? And therefore leaving subs holding stocks that the Port didn’t buy.
a) If the stock is suspended over the weekend, the Port may not have that info and the stock could still be recommended (of course it couldn’t be bought on Monday).
b) If a stock is illiquid the subs may submit a buy or sell order anyway. That would probably create at least some trades, and therefore, some volume, negating the “NOT be executed”.
c) If a stock is merged over the weekend then a sub not knowing that may submit an order anyway. That may or may not be filled depending on the specifics of how ALL the exchanges handle that.
Why give double weight to the closing price, when most of the commands are given before the opening, and not to give any weight to the opening price?
(Much more logic to use double weight to the open price)
Even if you choose to use an odd formula, it is important to give the designers option to use it in simulation
I think it is very important to use the same system before and after launch
Denny, the only way the R2G gets out of synch with real accounts is with stocks that are delisted due to acquisition , merger or spinoffs. What we will have to do is create a process to update the R2G and alert the subs.
In the case of spinoffs we just have to alert the subs to sell the new stock as this is what P123 does with cash equivalent dividends
In the case of acquisition for cash or merger we need to force the sell and alert the subs. Most likely subs did not get cash or stock in their accounts yet on Monday
Amiran, R2G are a way for someone to easily follow buy/sell of a strategy. Using open prices requires special commitment & trading skills, like placing limit orders, changing them throughout the day, etc, not what R2G is supposed to be.
We should probably disable allowing the use of open prices for sims that are used to create R2G’s
In the new ready 2 go page
In the “Trader” it show trading skill “High”
If people can’t put a simple limit order themselves, then they better not use a ready to go models at all
And with the more liquid models, sure the open price will be the execute price
I not see why the close price even relevant because all the orders sure will be fills before that price
But much more important than the logic you use, is to let us try the same logic in simulations
Marco - I’m not sure how many times this has to be stated but it apparently needs to be repeated again. There are plenty of R2G systems that have sufficient liquidity to handle market orders. The fact that low liquidity models and over-subscribed models are being offered does not mean you should cater to those exclusively. The open price is the generally the best option provided the model can support it.
The next point I would like to make is that your previous logic is backwards. You said:
“R2G are a way for someone to easily follow buy/sell of a strategy. Using open prices requires special commitment & trading skills, like placing limit orders, changing them throughout the day, etc, not what R2G is supposed to be.”
Many of us don’t use IB and don’t have the ability to do VWAP. Thus use of some variant of (O+H+L+C)/4 really means some form of special commitment & trading skills.
In contrast, use of market order and high liquidity R2G models is very easy for anyone to follow. I have been using market orders at open for many years and you know someone forgot to tell me it was difficult or requires a special commitment.
The fact that you have to get up early is the case no matter what orders you use to execute your trades, whether it be VWAP, limit or at market. And the problem lies with the fact that you went to Compustat. Otherwise we could be placing our orders the night before.
“R2G execution prices should be what the average sub could achieve, so vwap is probably the best”
Have you done a poll to see whether the average subscriber has the ability to do VWAP? How much work has been done to show that VWAP is equivalent to the formula you propose?
And if you choose to go with such a formula, Amiran is correct in the (2*O+H+L)/4 is a better option.
Tend to agree with Amiran that ‘open’ prices should be used in some way in the ‘formula’ figuring estimated model fills. For anything in the ‘trader’ or ‘lower liquidity’ category, some form of limit order prior to market open is likely the most basic form of order I would expect most P123 traders with hi-turn systems are using.
But…I’m fine if P123 doesn’t allow them. But, developers / P123 users ideally need the ability to build sim’s with any proposed ‘fill price’ method before live release.
In my case, it’s not likely a huge deal. But in developing systems, I will change the turnover by altering ‘sell rules’ based on a comparison of a) factor duration / fading vs. b) slippage and transaction costs. If the perceived benefits of higher turnover outweigh the likely costs by some ‘margin of safety’, I will generally choose the higher turn. So…assumed fill and slippage prices matter significantly in some ‘sell rule’ decisions on certain types of systems (2 of the R2G’s I’ve launched to date). So…as a general policy, it’s probably better for P123 to allow developers the tools to backtest any / all potentially significant proposed ‘constraints’ before rolling them out live.
If we are confined to one approach I would choose (O + H + L + C) / 4. That gives you the two price extremes of the day, the two time extremes of the day, and four different prices you know someone received a fill on. It doesn’t give you VWAP, but over long periods of testing it is going to be good enough. Not all traders have algos and will be able to approximate VWAP, but anyone trading size can do some form of TWAP. And it balances the perspectives that you see on the board. And doesn’t require more data. One’s perspective is understandably influenced by the size you trade (institutional vs high worth vs small trader) and by the volume/liquidity of the issues you are trading.