Last edited:
at close tonight i will post the running total results for this strat for the last 10 days. and also the results if trades were reversed. (i.e. same position sizing rules only if stock is +5% at 19:00 go long, if stock is -5% at 19:00 go short.
Glen
Glen
tonights trades closed
View attachment 31652
10 days, 60 trades, unsuccessful so far.
post results later
Last edited:
Running totals (Day 10)
Running totals sheet attached below (10 days, 1 tab per day)
P/L for tonight +$500
account balance =$27,103
View attachment 31653
Running totals sheet attached below (10 days, 1 tab per day)
P/L for tonight +$500
account balance =$27,103
View attachment 31653
Last edited:
Could the reverse of this strat be a good strat in the making
ie. if i had of gone long on the +5% stocks & short on the -5% stock, i would have been up the guts of $3K within the 10days. If i had of placed stops could have been more
i will continue to test this strat and the reverse of this strat next week..
also i've now got hyperserver 3 for TWS and TS2i (just got to get it to work now!)
thanks for the help guys
Glen
belflan,
With the strategy mentioned in my post at the start of the thread, I found that the stock selection is quite important, perhaps critical. Certain types of stock are more likely to exhibit the mean reverting behavior that you are looking for. Perhaps Grey1 might like to comment on this. I know that Grey1 has warned about the difficulties of trading fast moving stocks driven by high volume algo trading. Maybe we are talking about stocks like RIMM in this category.
If we are looking for a reversion to VWAP, then I would guess that we should be looking at large cap, high volume stocks with moderate (not low) volatility. Probably with large institutional participation (by institutional, I mean mutual funds, pension funds etc rather than hedge hogs). I would guess that VWAP directed traded volume might be relatively high is such stocks.
Of course this is all supposition, and I might be well off the mark. So we can put our powers of reasoning (which might not be all that reliable) aside and do it empirically. That is mostly what I did. I looked at various industries, sectors and other groupings of stocks and found that within the parameters of the system mentioned in the first post, the stocks of the DOW 30 are some of the best to apply that strategy to. It also seems to work on pharma. I also tried the 100 highest volume NYSE stocks and selected the most suitable. It does not work well on the NDX or biotech stocks.
There was no "future knowledge" used in deriving the system metioned in the first post. The stocks were chosen from how they behaved over the prior 6 months and the posted equity curve is strictly walk forward.
Average net profit on that system is about $0.13 per share per trade, and somebody asked about max DD which is about 10%.
My objective is to develop it further into a system that can be auto traded, for the very good reason that I live in Australia and US market hours are just horrible. I have a software development background, so the automation will take work, but is very doable. This type of system which is at least partially hedged a lot of the time has the advantage that some type of system failure - most probably the internet connection - can occur without incurring a catastrophic outcome. You can automatically place a MOC order as soon as a position is opened to ensure that you are not carrying positions overnight and programmatically cancel it if a postion is closed for some other reason.
I have some other comments about risk management, but I will leave them to another post.
With the strategy mentioned in my post at the start of the thread, I found that the stock selection is quite important, perhaps critical. Certain types of stock are more likely to exhibit the mean reverting behavior that you are looking for. Perhaps Grey1 might like to comment on this. I know that Grey1 has warned about the difficulties of trading fast moving stocks driven by high volume algo trading. Maybe we are talking about stocks like RIMM in this category.
If we are looking for a reversion to VWAP, then I would guess that we should be looking at large cap, high volume stocks with moderate (not low) volatility. Probably with large institutional participation (by institutional, I mean mutual funds, pension funds etc rather than hedge hogs). I would guess that VWAP directed traded volume might be relatively high is such stocks.
Of course this is all supposition, and I might be well off the mark. So we can put our powers of reasoning (which might not be all that reliable) aside and do it empirically. That is mostly what I did. I looked at various industries, sectors and other groupings of stocks and found that within the parameters of the system mentioned in the first post, the stocks of the DOW 30 are some of the best to apply that strategy to. It also seems to work on pharma. I also tried the 100 highest volume NYSE stocks and selected the most suitable. It does not work well on the NDX or biotech stocks.
There was no "future knowledge" used in deriving the system metioned in the first post. The stocks were chosen from how they behaved over the prior 6 months and the posted equity curve is strictly walk forward.
Average net profit on that system is about $0.13 per share per trade, and somebody asked about max DD which is about 10%.
My objective is to develop it further into a system that can be auto traded, for the very good reason that I live in Australia and US market hours are just horrible. I have a software development background, so the automation will take work, but is very doable. This type of system which is at least partially hedged a lot of the time has the advantage that some type of system failure - most probably the internet connection - can occur without incurring a catastrophic outcome. You can automatically place a MOC order as soon as a position is opened to ensure that you are not carrying positions overnight and programmatically cancel it if a postion is closed for some other reason.
I have some other comments about risk management, but I will leave them to another post.
Last edited:
Thanks for your post dcraig (very informative)
I'm reluctant to read to much into 60 trades, but maybe i need to additional filters to make this work (ie to pick the stocks more likely to mean revert, perhas some one help with this)
if i could seperate out stock that show mean reverting bebavior i could forward test
1.)mean reverting stock, short +5% stocks at 19:00 uk time, long -5% at 19:00uk time
&
2.) do the opositise for stocks that show bebavior of continuing with the trend (i probibly need to read Grey's path of least resistance thread.)
How could i draw up a list of such stocks? (is there an easy way?)
would you auto trade this this TS or build your own
look forward to reading it
thanks again
Glen
I'm reluctant to read to much into 60 trades, but maybe i need to additional filters to make this work (ie to pick the stocks more likely to mean revert, perhas some one help with this)
if i could seperate out stock that show mean reverting bebavior i could forward test
1.)mean reverting stock, short +5% stocks at 19:00 uk time, long -5% at 19:00uk time
&
2.) do the opositise for stocks that show bebavior of continuing with the trend (i probibly need to read Grey's path of least resistance thread.)
How could i draw up a list of such stocks? (is there an easy way?)
would you auto trade this this TS or build your own
look forward to reading it
thanks again
Glen
For screening stocks, you might like to have a look at StockFetcher.com - Stock Screening It's quite a powerfull technical screener, easy to use and very cheap. US stocks only.
I'm in the process of developing my own screener in Java. It is both technical and fundamental though the fundamental side is not yet as developed as the technical side. Currently my data sources are Yahoo for EOD and primarily OpenTick for intraday historical with support also for IB TWS historical data. I'm screen scraping fundamentals from Yahoo for US stocks and looking around for a better source for fundamentals for ASX stocks. If anybody knows of a free source of EOD historical data and/or fundamental data for Japanese stocks I would like to hear about it. All of this is a real PITA and I look forward to the day when the cost of a real professional data feed with a JAVA API is justifiable. And by professional, I don't mean eSignals crummy desktop API and similar such stuff.
This is a lot of work and I wouldn't recommend it to anybody. One reason I chose to do it is, as I said, I'm in Australia and I don't think trading US markets is sustainable in the long term because of the time zone except perhaps by autotrading. You then have the problem of tools and data sources and the picture is not nearly as good as for the US markets. If I build my own, the tools side of the problem is solved and I remain very much in control of my own destiny. If you roll your own, there is never anything you can't do because of software restrictions. Given sufficient work you can always develop the tool to requirements.
So no, I'm not using TS or RadarScreen, though I already have a fully functioning real time scanner written in Java that will do RadarScreen type stuff (and with much higher performance than RadarScreen, I might add).
I'm sorry I don't really have a good suggestion as to how to backtest other than use TS or maybe WealthLab and write the scripts. In the end it always comes down to putting in the work.
I'm in the process of developing my own screener in Java. It is both technical and fundamental though the fundamental side is not yet as developed as the technical side. Currently my data sources are Yahoo for EOD and primarily OpenTick for intraday historical with support also for IB TWS historical data. I'm screen scraping fundamentals from Yahoo for US stocks and looking around for a better source for fundamentals for ASX stocks. If anybody knows of a free source of EOD historical data and/or fundamental data for Japanese stocks I would like to hear about it. All of this is a real PITA and I look forward to the day when the cost of a real professional data feed with a JAVA API is justifiable. And by professional, I don't mean eSignals crummy desktop API and similar such stuff.
This is a lot of work and I wouldn't recommend it to anybody. One reason I chose to do it is, as I said, I'm in Australia and I don't think trading US markets is sustainable in the long term because of the time zone except perhaps by autotrading. You then have the problem of tools and data sources and the picture is not nearly as good as for the US markets. If I build my own, the tools side of the problem is solved and I remain very much in control of my own destiny. If you roll your own, there is never anything you can't do because of software restrictions. Given sufficient work you can always develop the tool to requirements.
So no, I'm not using TS or RadarScreen, though I already have a fully functioning real time scanner written in Java that will do RadarScreen type stuff (and with much higher performance than RadarScreen, I might add).
I'm sorry I don't really have a good suggestion as to how to backtest other than use TS or maybe WealthLab and write the scripts. In the end it always comes down to putting in the work.
Glenn
Experienced member
- Messages
- 1,040
- Likes
- 118
Using Sector Rotation is one way to identify areas where money is likely to be flowing, in or out. This also makes sense in the context of VWAP and large scale institutional positioning.
With Radar Screen is it possible to set up analysis of all the Sector Indices and then periodically check using Daily/Weekly MACCI to determine likely candidates. This does not have to be running all the time and using up cpu resource and memory, but instead have a TS workspace containing the Radar indicators and periodically paste the tickers in to see the picture (e.g. weekly or monthly).
Once you have identified a Strong or Weak (i.e. OB or OS) sector you can then scan that sector for the stocks which fit the parameters for price, volume and volty.
Glenn
Thanks Glenn. Yes that certainly sounds like a decent approach. I have been thinking along somewhat similar lines, at least in terms of sector rotation, but I have been spending so much time and effort doing software development that I havn't had much opportunity to do the market analysis work. I don't need TS or RadarScreen. My own screener/scanner has sector/industry analysis capabilities to do this sort of thing.
Last edited:
Last edited:
Running totals (day 11)
another poor day for the strat
P/L on day=
-$825
Account Balance = $26,278
reversing entries would have been better again tonight
View attachment 31703
another poor day for the strat
P/L on day=
-$825
Account Balance = $26,278
reversing entries would have been better again tonight
View attachment 31703
Last edited:
kako
Active member
- Messages
- 125
- Likes
- 4
Hi
Found this article lately:
http://www.itg.com/news_events/papers/TP_Spring_2002_Madhavan.pdf
and thought it might be useful to ppl concerned with vwap strats as we all are
Belfan, keep up the good work. I am interested in what you're doing and posting in this bb.
Regards
.......................
Kako
Found this article lately:
http://www.itg.com/news_events/papers/TP_Spring_2002_Madhavan.pdf
and thought it might be useful to ppl concerned with vwap strats as we all are
Belfan, keep up the good work. I am interested in what you're doing and posting in this bb.
Regards
.......................
Kako
hi kako,
i too been looking at vwap articles on the web............found this interesting site Wilmott Forums - VWAP Engine Available for free?
seems that some brokers and traders use the previous close as their benchmark instead of the vwap.
jason
i too been looking at vwap articles on the web............found this interesting site Wilmott Forums - VWAP Engine Available for free?
seems that some brokers and traders use the previous close as their benchmark instead of the vwap.
jason
OK, here's another link to a forum discussing VWAP bands based on 1 and 2 standard deviations of price from VWAP. The chart of YM on the third page of the thread is quite interesting, but I wouln't draw any conclusions without a much closer look.
VWAP Indicator with 1SD and 2SD bands - Page 3 - Traders Laboratory
VWAP Indicator with 1SD and 2SD bands - Page 3 - Traders Laboratory
great link there dcraig,
interesting to see that traders on that forum are using vwap in trading futures markets such as er2, I was not aware that vwap could work in futures markets.
Also they seem to be using the vwap in a different way to grey........they wait for the price to touch vwap before trading and take profits at 1st and 2nd deviations.....I wonder why they do this?......I would have thought buying or selling at the extremes of the vwap price would be best and then exiting once price touches vwap.
I'm going to read through the whole threads there on vwap, maybe i can gain more insight..............thanks again for the link.
jason
interesting to see that traders on that forum are using vwap in trading futures markets such as er2, I was not aware that vwap could work in futures markets.
Also they seem to be using the vwap in a different way to grey........they wait for the price to touch vwap before trading and take profits at 1st and 2nd deviations.....I wonder why they do this?......I would have thought buying or selling at the extremes of the vwap price would be best and then exiting once price touches vwap.
I'm going to read through the whole threads there on vwap, maybe i can gain more insight..............thanks again for the link.
jason
Last edited:
Last edited:
Last edited:
