peterbee77

Far be it for a beginner like me to challenge the wisdom of the masters, but here goes. A way of keeping faith with the KISS concept would be to stringently test any rule proposed to be added, before adding it. Siuya's and Mystic's rule proposes adding a test of trend to the system, being the 50/200 SMA crossover. However, the original system already contains a trend test, implicitly - the 2h box itself. It defines the trend with reference to a breakout beyond the 2h box's High and Low. As today's chart in the above post shows, the box's intrinsic test of trend might be a better test because it is current, rather than looking back too far. Also it is amenable to smaller trend changes, whereas a 200MA has a lot of inertia and will "miss" a lot of a new trend development. Maybe the proposed MAs are too slow for the limited stop-loss / take-profit moves proposed in this system? Hence I propose that the 50/200 MA test remain a "probationary" rule addition at best - to only be added if it can be shown to be better than the simple box trend test. Btw, thanks for starting this thread Mystic. I'm looking forward with much interest to see where it goes.

"Don't be afraid to buy high and sell low" I include this saying because I thought I'd misread it first time. It comes from an article entitled "Flow of the Markets" by Van Tharp, at his website, on the page entitled "Free". I thought the article worthwhile, probably because it relates to one of the problems I have trading.

'Who do you blame when you sell out of your profitable positions too early because you are afraid they will pull back and turn into losses? Or you hang on to your losing positions as they ratchet down, each little up-tick giving you hope, but finally sell out at the bottom because the pain is too great? This syndrome is described as "eating like a bird and pooping like an elephant". ' S. Bigalow, in Profitable Candlestick Trading (2002)

"One of the most helpful things that anybody can learn is to give up trying to catch the last eighth or the first. These two are the most expensive eighths in the world." Jesse Livermore in "Reminiscences of a Stock Operator"

"Throughout all my years of investing I've found that the big money was never made in the buying or the selling. The big money was made in the waiting." Jesse Livermore

If you'll pardon my obtuseness, I'm puzzled as to why you would have gone long the USDCHF trade (spot) in mid-January 2010, per the chart? When I look at the immediate chart only - without looking into the future off to the right - I see a medium strength countertrend rally in Dec.09 which caused the fast-MA to cross only slightly above the slow-MA before fading. The price turned back down in late-Dec and crossed back below the EMA65, with the EMA21 also turning down a few days later - which would have suggested to me that the previous downtrend was re-asserting itself. By the time the sRSI rose above 0 and gave a Long trigger signal on Jan.12, the fast MA had crossed down, which would have suggested to me to go short (or do nothing at all until the trend became clear) rather than going long at that point in time. That's how I'd interpret the basic direction-setting rule of the MAs in the instructions at the beginning of this thread. It looks like a case of departing from the rules, based on other factors, and it would be instructive to hear the reasoning behind it. Looking forward to your thoughts OT, with thanks, Peter

Hi Neo, I must admit I'm a fan of my charting software, which I've been using for 7 years now - I'm still learning new features it has. It's sold by a little Aussie company called BullCharts, which suits me because I trade Aussie stocks. Their website is " http://www.bullcharts.com.au " and has a bit of info about it. Even though it's primarily designed for Aussie stocks, BullCharts will also accept other data in Metastock format, which is an industry standard data format I believe. So you might be able to run your Ninja daily data feed through it (I'm assuming you live in the US of A). I'm not sure how real-time data would work - it works fine with Aussie data. Briefly BullCharts comes with about >250 inbuilt indicators (including Stochastic RSI), each of which can be tweaked to your own preferred parameters. But what I like most is you can code your own indicators, or variations of other's indicators - do-able by someone like me, who is not an IT professional. Their stock scanning is menu-driven so you can easily construct a scan that applies any of their indicators to a number of stock fields (open, low,close, high, volume, value-traded, no. of trades, median price, etc), or you can also include scan criteria based on pieces of your own-coded script. One feature I've found useful is that some scanning criteria can be set to "must be met" while others are set to "optional" and you assign a weighting to each of them, so you end up with a list of stocks ranked according to their total scores on your weighted optional criteria. After looking at their site you might want to talk to the head guy, Brendon, as to whether it might be feasible to adapt it to your needs. Note their phone no is local Australian, so you'll need to delete the leading zero, and add your international access code (is it "00" in USA?) and Australia's international dial-code ("61") to the phone number on the website (so "(02) 98.." becomes "00 61 2 98..". And bear in mind we're 14 hours ahead of USA Daylight Saving time at the moment, so 8:00pm Thursday USA East coast Daylight Saving Time translates to 10:00am Friday Sydney time, for example. Good luck, Peter

Thanks O/Timer, for your helpful thoughts in post #151 (page 19) responding to my post #146, re applying your method to stocks. Your comments have crystallised some trends I sensed in my preliminary backtesting. Applying your method to stocks, one really does need to filter out A LOT of stocks, as you suggest. Luckily my scanning software allows detailed filtering - my scan selected for stocks meeting the following criteria: EMA65>EMA21 (ie I focussed on long trades only); price above EMA65; StochRSI<50% (ie approaching oversold); filter for higher beta; filter for high average daily liquidity. This gave me a “shortlist” of 80 stocks out of the 2,500 Oz stocks which I followed until they triggered . I will look at incorporating some fundamental scan criteria as well, maybe similar to the IBD-50 (which applies the CANSLIM criteria I think). One thing I didn’t mention in my earlier post was, I am using a random-number generator to select dates for my backtesting, and the first date it selected was Feb-18-2011. When I looked at a chart of the indexes today (see thumbnails), I see that some of my "shortlist" did OK relatively. Feb-18 was near an intermediate top: it was followed by a mild decline; then a levelling out (during which time half of my shortlist triggered an entry under the OT criteria); followed by a steep decline into the mid-March bottom. Actually the charts of many stocks were reminiscent of your chart of the AD contract (AUDUSD) in your post #75 on page 10. There, the EMAs met the entry criteria on 9 occasions in that the fast-EMA was above the slow-EMA the whole time, and the sRSI rose from oversold, and the price action triggered an entry. However, as you stated the price kept on steadily making lower lows and lower highs. The parallels between my stock charts and the AD arises from the fact that the Ozzie (I'm using the phonetic spelling rather than "Aussie") is a commodity currency and when there is a “risk-off” mood then many assets sell off, whether they be the "risk" commodities, commodity currencies or stocks (global stocks underwent similar declines between April and June). Question is, is there a way of translating that risk-on/risk-off mood into technical analytic terms? As you say OT, "This situation will have to be addressed." Any ideas, anyone? cheers, PB

I must say I've greatly appreciated this thread - many thanks OT! I'll apologise at the outset for raising a side-branch to the main thread here. I'm interested in applying OT's method to stocks (even worse, Ozzie stocks) and I suspect they behave differently to commodities or forex. I've come across a problem while ploughing through a series of backtests of Oz stocks, and I was prompted to raise it when I read the "exogenous" thread quoted by un$ane in post #141 (thanks un$ane): I was struck by one of the rules in CeazaredeBeer's method: CeasaredeBeer restricts the range of the Signal candles he takes (ignore the $5 range size he mentioned in the quote - I'm taking it out of context - it's the principle that's interesting). The problem I've come across with Oz stocks (maybe all stocks) is that sometimes the short-term correction against the main trend is reversed by a strong bar that reverses most of the countertrend move in one bar. When this large-range bar occurs during the overbought/oversold phase of the correction it becomes the Signal bar, by definition per OT's method. Going long for example, the stochRSI-7 may go from 0 to 70 on the first day. The price then has to rise above that tall Signal bar to trigger the Entry. But by the time it's risen that far it's fairly close to oversold and the trend might pull back soon after. Result = another entry followed by a whipsaw loss. I'm ploughing through backtests of OT's method in stocks, and I've come across the above pattern a few times amongst the failed trades. I've just started the backtesting, so I don't know yet if it's enough of a problem to worry about. But I was wondering whether I shouldn't formulate an additional rule, say, to reject OB/OS reversal Signals that swing from sRSI= 0/100 past 50 in one day. I know one shouldn't complicate a good method with extra rules - and OTs method is certainly great in that respect. But it might be useful for stocks. PS: I can sympathise with Ingot's earlier posts where his preferred entry was to enter early on in the day when sRSI made its move - rather than waiting for sRSI to close and wait for the next day to enter, thereby collecting himself a goodly number of extra pips more than the per-rules entry. A reflection of the same problem maybe?