RichardTodd

Maybe I'm an idiot, but this all seems very straightforward to me. If people find the indicators and methods of value, and some % of them either get into his FX/IB deal or send donations, then that is TRO receiving compensation for providing value. End of story. It's just basic business concepts, as far as I can tell. TRO draining the banks or not is irrelevant. Let's say he made $20million this year trading. Does that have any bearing on your ability to make $20million this year? People have made fortunes with a minute chart and an EMA... and since we all have those I guess we should all be rich, right? On the other hand, if he lost $20million, does that mean the indicators are useless, or that his broker deal is sleazy, or that donations are not warranted when people like what he's done? Not in any objective sense. It's one data point, people, and he may or may not even use the indicators anyway. Indicators only indicate. It's the trader that trades. As another trader with a CS degree and a love for programming and passive income, I can tell you that there's nothing suspicious about prolific coding. When I'm active, I'll produce 3 to 5 indicators a night. Why? It's fun to do, no other reason. I also often take code people send me or that I stumble across, and make various improvements. I do it just for the heck of it. Like, I guess it's why some people have fun working on cars, or participating in sporting events, or whatever.

Well said. In my view, when people say they've sworn off indicators, they usually mean they've sworn off lowpass filters. Many of them still use bars and all kinds of further derived info like trendlines and fibs and pivots. And to me, any derived information not on the tape itself is just as much an indicator as age-old oscillators.

Yeah, I used an alternate method that's eager to jump down and a bit more reluctant to jump up. The reason is that there's not much penalty for including too many traders for a bar or two (just a bit more noise), but including too few traders leaves the line flat. That's funny you should say that, because the dynamic splitter I made had both a minimum and a maximum that you could set. They were just static inputs, rather than SD-style outlier detection, though. In other words, no matter how slow the market was, I didn't want my splitter telling me that 5 contract trades were "big." I put in the maximum thinking of those extraordinary blocks you see in the tape on stocks sometimes, which in my experience has the opposite effect of a normal block if it has any effect at all. I actually didn't know about off-market globex trades. I don't think I've ever noticed an obvious one, so that's very interesting. Maybe I don't know how to spot them.

Well, if you consider how a macd works, then you know the line flattens out because the ema's are still separating but at a slower rate than before. So, momentum is coming out of the underlying in those cases, but the underlying isn't pushing down. Price coming down during an uptrend without big traders bothering to sell == potential for profit$$$

Actually I did this last summer when it annoyed me that sometimes during lunch there were no 100 contract traders at all. The version I did can trade on a single chart from the day session into the evening session and it transitions over a few bars to the new environment(s). While it's neat that it can do that, I wouldn't say it's surprisingly powerful. "Wonderfully convenient" is more like the extent of it. I was actually a little disappointed. It does seem to take a little noise out of the line when really big players come to the game (since it starts focusing only on them), but on fast charts those same really big players will be so much of the volume that they will dominate the shape of the splitter line anyway. So, I didn't even bother to release it (because I was too lazy to make the enhancement for TS, eSignal, and our other platforms) (I did it on ninja). But, lately it's been pointed out that people have trouble finding the 'right' settings for their splitter on other markets besides ES, and the dynamic splitter would 'solve' that problem for the common case where you just want to follow the big players. So, sometime in the next month my company wants me to release it after all. The new version will have a few enhancements to how it judges the line thickness as well.

Maybe you can use 25+ sized ticks well (on ES I assume), and maybe not. All I can say is that there is no shortage of 200+ sized ticks, and 25 contracts are obviously NOTHING to the bigger players. In other words, if you see lots of 25 contract market sales and you think "well, they know more than I do so I should sell too"... then you are blatantly in error from what I can tell. Traders who trade in the 1000's could frankly be doing just about anything in 25 contract increments. To put it another way, I don't think those trades carry much information, in an information-theory sense. It's just noise on your signal. I know I sound like a broken record, but no one seems to hear me when I say: the main function of the indicator is to bring out times when large traders are in a big enough hurry to show their hands with large market orders. A secondary use of the indicator is to show that the small orders tend to be against the direction of the eventual move. All of the other stuff people want to believe they can see might be in there, but is probably better shown by another (possibly related) indicator. There are plenty of times every day when, if you understand this, the indicator can help you find high-quality entry points. To find them, you have to think about what the line means, compared to what price means. I pointed out one very simple strategy a few posts ago against the macd version on a 610-tick chart posted here. There are other tradable ideas, and in some conditions the line will turn ahead of the move, sure enough. But it's not a magic line that always leads the market. If you are looking for that, you will have "issues" with every indicator you find out there. I know I probably sound harsh, and I don't claim to know everything about how to use the indicator, but I do wish people would start from what the indicator means, and not blindly look for divergences or leading action or whatever. I would say the same thing about any indicator.

Well... we implement stuff for ourselves, and then share it with our clients when we stumble onto stuff we like. So, even if something might be worth implementing, unless one of us wants it for ourselves, it won't get implemented. None of us trade minute charts, so we wouldn't have tried out SV4 if it weren't for multicharts. Our TS customers do get the code, and they can indeed run it on minute charts to the best of my knowledge. But like you said, we don't go out of our way to advertise it because the questions would be endless. Also that video must be over a year old now, and be aware that none of us at EOT uses SV4 anymore. I honestly don't remember why we quit (maybe just because ER2 went to TF and we were too lazy to reconfigure!), so maybe an investigation would renew interest. Sometimes we have so many ideas going that some get lost along the way. BTW if you did try to clone SV4, you'd quickly find that you have to go a completely different way than your close-comparison solution. So, it won't be just a matter of making the script "longer" as someone put it. Multi-symbol indicators going tick-by-tick are still tied to the primary symbol to which you apply them... so even though you have access to data1 data2 etc, you'd only be executing on each of data1's ticks. So you can't detect data2's big trades the way you are currently doing it. If you don't understand what this paragraph means, then don't even try it until you've built a few simpler multi-symbol indicators. It takes some getting used to.

I meant to also say: another easy way to 'stress-test' models is to put them on, say, 800+ trades instead of 100ish. This way, you don't have to check as many data points, and the biggest trades are arguably the most important to get right anyway. In other words, if I had a model that wasn't as good as 4 at catching 100-lots, but ALWAYS got the 1000-lots correct, that's a trade I'd take (no pun intended).

A line that goes straight up would appear to anticipate every up move, so I'm not sure that's the best way to go forward. If you want suggestions, I'd probably take a screen capture movie of the ohlc (non-macd) of the models beside a time and sales window, and review it for anomalies. If there are easy ways to improve upon the models, that'd be how I went about finding them.

I think the main difference will be like the cases I pointed out in the posted 610 chart, which are important to me because I trade off of them. Doesn't appear to be as important for people looking for swing high divergences and such. But, when price goes up and big traders sell into it, v2 will see some of those orders as buys, which is why it tends to turn up with the price.

Well I'm assuming v4 was in red, right? (I just hope it's not the middle one!) :-) Anyway, if you look at the red line on the 610 chart in particular, you can see that just before 9:30, just before 10:16, and just before 11:04, and right around 12:48 and 12:59 are places that price pulls back but the splitter doesn't turn. This indication that the big boys and girls are fading the pullback is one of my primary entry techniques as long as it's with the trend. I've gone over this simple technique a couple times with my live room folks, and it's always popular.

How funny.. I thought you were your brother! Yes we've been having a lot of fun watching the time and sales with him to pinpoint entries. He's had some valuable insights.

I think in general you'll be happy that you did. When I look at the time and sales on the eminis, it is rare when the bid and ask rises and a big trade hits the new bid before any trades hit the new ask. I'm sure occasionally this happens and the heuristic gets it wrong, but if the alternative is designing it so that the trade is ALWAYS missed, I'll take the heuristic any day of the week. Not to mention, being able to spot a case where several big orders hit the same price in a row, and measuring it properly, will make the splitter more accurate in the case it was designed to detect in the first place.

think of a case like: 901 1 900.75 1 901 1 901 800 901 1 will version3 catch the 800 trade? Glancing at it, I think it will not, because close = lastclose. I had given pseudo-code that I think works for this case, previously, though I'm sure there's any number of ways you can fix it. Another thing, for enterprising people that want to build from here, is that lots of posts ago, Blowfish had pointed out an issue that most indicators like this have. The first tick they see can be incorrectly judged as a large trade because of pre-existing volume that had built up in a bar. To properly avoid this issue, you have to ignore the first tick you come across, aside from accounting purposes. At least, that's one way around it. On a fast chart, this is a small issue. On a 17711 share bar chart, it sucks when your splitter thinks there was a 17000 contract trade on the first tick it sees.

Ok, fair enough. Given that buying 800 contracts at 901 and then selling them at 901.50 looks like 1600 contracts bought to macd2, I'd suggest the first way folks build from there is to change that part of the model. If my commercial contracts allowed it, I'd write you some code to play with. As it is, the pseudocode I've provided is already more than I should do. I'm afraid I can only help so much. I'll shut up now about it. I'm just trying to help things along.

You guess. I would suggest you assume that an uptick followed by a zero tick is two buys, because when you watch the tape this is statistically far more common. If you update your guess on every trade as in the pseudocode I gave, and not just the big trades, I expect that you will rarely be wrong on the eminis. The biggest problem with only guessing the up vs. downtick on big trades is that for certain settings, you'll only get a big trade every few minutes. Price can move a lot in between... Say someone buys 800 contracts at 903. Then five minutes later someone buys 900 contracts at 902.50. Well... if I read it correctly, the algorithm in the macd2 indicator I saw would call the second trade a sell because just because it was lower than the last block trade. I feel it's much more accurate to keep track of the best guess upticks vs. downticks on every trade, and then only count the big ones. It's only a suggestion, of course. I'm quite happy with the splitter I have, so...

Yeah, I wasn't trying to be novel or "provocative"

Exactly, though to be fair all the indicators I've seen here have grouped the data into bar-sized groups and then averaged/macd'd/etc them. Still, I'm not sure why this particular thread was hijacked to talk about the usefulness of statistical sampling. I've done some indicators recently that operate on the last x ticks regardless of what bars I'm looking at. It's nice on some indicators to avoid the discontinuity at each fresh bar. Of course it makes the programming more tedious, and then the charting platform snapshots it at bar intervals as the chart progresses, so you are back to a sampled signal in the end. Anyway, I'd love to see threads about nonstandard chart types and ways to avoid sampling. I'm not sure what's so novel about nonparametric studies but even that might be interesting. Just not sure why it's here in a thread about a volume study.

What's the advantage of ignoring the subsequent block trades? If someone hammers away at a price level 100's of contracts at a time, I can't think of a reason off-hand that I wouldn't want to count them all.

Sorry I'm late to this party, I'm just now looking at this thread. It makes me think: if my assumption is that a market player is 'pulsing' the market with orders to hide their considerable volume, then: 1) Intensity would be high (contracts/time) 2) total pulsed volume would be high (otherwise who cares?) 3) total pulsed volume would be largely directional (mostly upticks or mostly downticks) (doesn't that seem sensible, or am I missing the point?)... so any algorithm I'd devise would combine the 3 factors. so maybe... volume/time is intensity... and if I post-process the 'pulsy' trades, then: some variation of volume^2/time would highlight high volume 'pulses' over low volume ones... and if I change it to (volume*(upvol-downvol))/time it would be similar to volume^2/time in amplitude if the pulse is highly directional, and diminished if it is a mix of buys and sells. I don't know, just thinking out loud. I may just try it, sometime. (and yeah, it sucks that TS doesn't split trades for their volume bars like the rest of the world... it causes our users all kinds of "my chart doesn't look like your ninja chart" anxiety)

hmmm.. if I'm reading this correctly then this code will ignore the second of two block trades in a row at the same price? And it decides up vs down by comparing each block trade to the last block trade... I would think if you are going down this road that it'd be better to guess upticks vs downticks on every trade, and then only count the ones that happen to be big blocks... something along these (pseudo-code) lines would be my suggestion... hopefully it is clear enough: intrabarpersist bool uptick(true); // guess that close above last close is // an uptick, and vice-versa if (close > lastclose) then uptick = true; if (close < lastclose) then uptick = false; // ... and implicitly if close == lastclose then uptick should still // be whatever it was last time price changed lastclose = close; // remember for next time if(trade is large) then begin if(uptick) count as upblock else count as downblock end You might want to try something like that, anyway. Actually at least on the e-minis it's probably a pretty good heuristic.

Oh, I think I know what it's really good at, but I've watched it every day for almost a year now. It takes time to see the nuances. My hair stands up a little when I see people interpret the splitters in ways I think are probably wrong... but I don't see any reason to get bent out of shape about it. They are making inferences from too few cases, plain and simple. Imagine how much more helpful it would be if you simply posted an example of what you'd consider good analysis. Since you seem very sure of yourself, I'm sure it would be of interest to all. More interest than essentially calling everyone an idiot, at any rate.

I agree (and have said more than once) that it doesn't do you any good to try to match our one implementation bar-for-bar. I would caution that what you probably want is a family of related indicators for various tasks. To say that any indicator is "accurate" or does "a better job" without context is a bit off the mark in my opinion. A moving average is always 100% "accurate" if it is calculated correctly, and nothing could do a "better job" at showing you the moving average.... it just might not be the best fit for a particular trading plan. For certain applications, our splitter, or phall's, might be perfect. For others, you might want to manipulate the data a different way. For tape reading/VSA-type stuff, I'd massage data a bit differently (in particular for starters I would try to bring out cases when traders seem to have trouble eating through price levels, as this would help you see the limit-order side of large supply and demand). In fact I'm starting to do statistical research on that very topic in my spare time.

I wouldn't worry too much about what I do, if I were you. I trade fast, on fast charts, and when I use the splitter, it is also turning very quickly. It's not set up to show me divergence. We have another, smoother, volume indicator called VR4x1 which I can use when I trade divergences off fast charts. That doesn't mean you can't use a splitter for divergence, as long as you give it a big enough time horizon.

Well, I'll just say... Never forget that you are only tracking the large market orders with the splitter. Of course big traders will also trade in small market orders and small/big limit orders. So, do not expect a whole story. I've tried to remind people several times that the main purpose of the splitter is to show when the big traders act like there is no time to spare, because otherwise there is value in hiding their actions. As a consequence, it will show you when they run stops, and it will show you when they push hard directionally, and it will show you when they eagerly attack highly liquid areas (presumably because they are racing to be the ones who get to avoid slippage). Since our eotpro splitter shows ebbs and flows in their commitment, we can also see when they kinda just stand aside. But, in my opinion, when people claim they can see accumulation and distribution via our volume splitter or market delta type approach, I suggest that they are misleading themselves, at least a little. I can't think of a reason why big traders would consistently broadcast accumulation and distribution in such an obvious fashion. Maybe they do, but I'm skeptical, and I certainly don't use our splitter that way. (not to mention, most splitters I've seen in forums have such fast averages on them that the horizon effect should cause their indicators to completely miss any lengthy accumulation/distribution on a fast chart, even if the big traders are being obvious about it). So, I'd never discourage anyone from looking into any trading angles they can imagine, but I want to remind people to consider the limitations of their algorithm. The human brain can always find a pattern if it wants to, even if there's not one there!