9/28 and 9/29 - BOT1 didn't take any trades. I didn't get around to re-optimizing over the weekend so I'm running with somewhat stale settings but it'll have to do for now. Ends up that it doesn't matter since it hasn't fired any entry signals yet this week anyways.
My Reading Price Charts Bar by Bar (Al Brooks) book arrived today. I'm looking forward to diving into this. I do most of my book reading on my lunch breaks so it may take me a while to get through.
I coded a few price action based strategies I thought looked interesting, but could never get more than 40% success rate. The current strategy looks for 2 up bars. If I get 2 in a row I go long. If I get some even bars (open = close) in between that doesn't reset the count. Same for short: if I get 2 consecutive down bars (excluding even bars) I go short. I tried filtering for going in the direction of the trend (based on JMA), didn't seem to help. Then I tried only entering when an ADX was > 20, didn't get any better. Using a double stochastics to scalp looks interesting but I didn't code it yet. I'm going to try playing with instead of 2 green bars, I get 2 bars with higher highs (regardless of the candle bodies), then go long. Or 2 bars of higher lows, etc. These sorts of strategies work when it trends but I have to risk 4 or 5 to make 3 and sometimes it enters just as the trend just is ending (I know, we've all been there trying to trade trends). And in chop you get murdered. I'll keep playing with it, combine multiple time frames, other sort of filters that might increase win rates. We'll see what I can grok from the Brooks book too.
I'm doing most of my testing on the ZB which doesn't move a lot, but I'm just trying to scalp 2-3 ticks since the tick value is so large. I'm using small time frames like 34, 55 or 89 tick charts since it can make sudden, bursty moves in larger time frames and then immediately reverse the other way.
9/30 - BOT1 took 2 trades, 2 losers. Blah. Most of my failures come from periods of small volatility then there is a sudden burst/breakout. I need to program a way to filter these out because these are not good counter-trend opportunities when it is blasting out of low volatility periods. On the contrary, these should be good breakout trades. The trick is algorithmically defining what low volatility is. ATR? Tight Bollinger bands?
Hi shodson , I like your journal . These are not counter trend trades at all in fact these have a very high likelyhood of succeeding . I sent a screenshot of how I take them under the right conditions .
First a trend is not in place ( no stair stepping ) , price must make a really hard move to the LOD or HOD and at least 2 or 3 bars in a row should achieve the LOD or HOD. This is like a news release or opening bell - event driven . If you do this against a trend you will probably get ran over . Then ...
When a bar makes a higher low , using my example , wait until price trades above that bar or better yet closes above that bar . If price proceeds to a new LOD just wait again for a bar to make a higher low .
Bearish volume MUST be declining to confirm bullishness . Try to grab a retracement back to the start of the decline .
Dont get discouraged and dont think these signals are any more risky or unlikely to work than more conventional signals . Look at the screenshot , thats a 2 point stop and a 9 point gain if you didnt get spooked out early . The bullish volume increase would hint towards riding it out .
10/9 - No trading since early last week. Just too busy with project deadlines and I turned BOT1 off because I wanted to make adjustments before running it some more. I'm still playing around with strategies when I get a chance. I'm still reading Al Brooks' Price Action book but it's slow-going as each page has a lot to absorb. But already I'm starting to see bar charts more clearly and getting better at predicting which direction the next bar will be based on previous bars. This will be very useful and I look forward to absorbing more information from this book. For fun I'll put up a chart, go back in time, look at the bar on the right of the screen and decide, bar-by-bar, if I want to be long or short at that moment in time, then hit the right arrow key to see the next bar and decide again. My accuracy is improving.
I also made a new friend from the forum and I'm helping him with his strategy development. It was his strategy, I'm just providing the C#/NinjaScript coding. Maybe we can build something together that will be profitable for both of us.
I developed my own bias indicator, inspired from EOT's Allas Average, that I'm enjoying using in filtering out my trades. It's just an average of swing highs and lows, similar to some other indys I've seen others use.
I'm thinking it's time to get up some early mornings and get manually trading again, at least on SIM. I think I'm an paralysis mode, making excuses why it's not worth trading since so far after all the work I've done I'm barely above break-even. However, most of my efforts have been in automated strategy development and strategy research and not in becoming a better discretionary trader. I wish I could take at least 1 week off, really focus full-time my efforts, both discretionary and automated, and get myself to the next level. But I know I'm going to have a lot of year-end expenses/taxes due and I need to make as much cash as I can in my work right now while I have plenty of client work to do before the recession enters into a double dip.
The following user says Thank You to shodson for this post:
Been going through a bunch of old issues of TASC (all the way back to Jan 2009) that I got behind in reading and have been playing with some of the indicators and systems published in there (with or without code provided by NT support), but they all semi-suck. So I decided to spend more time on coding good money management techniques into my strategies. Implemented scalpers with runners (scaling in), break-even the runner when scalp target is hit, etc.
Also, I've been experimenting with "grippage" (opposite of slippage, is there a better term other people use?). If I get a signal, I put a limit order 1 or 2 ticks in my favor. I don't always get filled if it moves away quickly, but it increases the odds of success if the trade is taken in chop or if price wiggles around a bit after I get a signal and lowers the chance of getting whipsawed. Plus I can decrease my stop loss sizes and increase my profit targets. See attached image with PBFSqueeze, a moderately useful scalping signal generator, in strategy analyzer. With 20% win rate I can still break even with 2 ticks of grippage and 10 tick PT and 6 tick SL. Of course, limit orders are more complicated to manage in a live market so I'll have to compensate for that as well as I start to roll this stuff out into SIM and then a live market, such as how long/how many bars do I want to keep an unfilled order open, canceling the order if the market condition changes, how to programatically handle partially filled orders, etc.
Also, I'm closing my Tradestation accounts (IRA and small personal account). I just don't use it and don't have the time and energy to learn it. I'm too busy playing with NT, and getting tired of getting charged a monthly fee for software I don't use. Going to move my IRA over to thinkorswim, which I'm more familiar with and their platform is free, and get back into EOD swing trading and options trading/writing, which is what I used to do before becoming a day-trading / NinjaScript / futures.io (formerly BMT) junkie.
The following user says Thank You to shodson for this post:
Never heard the term 'grippage' but it makes sense =) Nevertheless I do make use of positive slippage when possible, monitoring the time between range style bars is helpful here.
As you mention, the burden of handling limit orders is heavy, it was way harder than I expected to make it work right in realtime.. but it is worth the effort for sure. There are some very cool and lucrative MM tricks that are impossible without the ability to quickly manipulate limit orders.
The way I handle no-fills is by assigning attributes to entry signals, ie each type of entry signal is assigned a grading scale for how powerful a signal it is, do we want to chase this signal if not filled immediately, if so how far to chase it, if not how long to wait before canceling order (can be #bars or time based).
I also use the concept of repulsive signals, ie after the signal the position where the limit order will actually be placed is 'pushed' by the price until some criteria is met. These are extremely useful for exits (dynamic target) but also entries as well (most likely to generate 'grippage' )
The following user says Thank You to sefstrat for this post:
How did you get your attached snapshot to show that upper pane having "Summary, Chart, Graphs, Execution, Trades, Periods, Orders, Settings" options. Some of these options are only available in the "Control Center" window dialogue. Do you have another version of the NinjaTrader ? I have version 6.5.1000.9.