Anyone have any idea where the data comes from for the symbol NISS-NYSE if you are not paying the 3.25 exchange fee for NYSE indices? I have been trying to switch over to SC for internals but there Tick and adv-dec leaves a lot to be desired. wasn't even expecting it to work since I had not subscribed to the exchange indices. but it does...
anyone else come up with a SC solution for internals?
Well for me it takes several mins for the adv dec reading to get to the level I see on other platforms. SC has advised that they feel theirs is more accurate. I don't share there same view. If the market opens up with a decent gap down like today. and your first reading is -216 and other platforms open at a -1150 that's a huge difference. I might be wrong but I believe that the more accurate reading based on a gap down and the logic that most stocks on there initial trade will be declining from there close based on balancing Cash to Futures as the cash market opens would be the -1150 reading. I have also found this too be the case with Ticks as well.
You are always going to see some variances between platforms they will never be exact but reading the strength of the opening is important for me.
The following user says Thank You to dougrkyle for this post:
I'll let the OP post his opinion, but I'll give a little perspective using the TICK as an example.
Most people use these internals because someone told them it was a good idea (and it is) and how to use them. This latter part is what gets people in trouble because vendor x's version of the internals does not look like vendor y, so obviously they are wrong. The issue is the person assumes that some version is "right".
When I started out, I used TradeStation and their TICK, and my first mentor also used this, and I got to know that certain levels (i.e. 400 and 800) were important for the method that was being used. TradeStation tends to be one of the gold standards for the TICK.
Later, I started using IQFeed, and had to adjust because their version just was not the same. It had the same shapes, peaks, and valleys, but the actual values were different. They claim to use the same algorithm, but have different values, hmmmm. No big deal though, because you just have to adjust your "magic" levels to the data that is provided.
Over time, I have been able to see versions from BarChart, ESignal, and a couple of others. Guess what, they are all different (even ones that are claimed to be sourced from the exchange). They all have generally the same shapes, but the values tend to be a little different. A big part of this is the sampling algorithm. If you were to calculate this index, you would have to look at all of the NYSE ticks as they go by and adjust the count accordingly. Typically, this is sampled anywhere from twice per second (Sierra) to once per 5 seconds. During that sample period, there are a lot of different values, and you can potentially miss the extremes. I know Sierra tracks the extremes during the sampling period (it is documented), not sure on the others. This is why one version might report 537 and another 564. This has gotten better as the sampling period has gotten shorter over time.
All of the vendors other than Sierra have one thing in common, namely that they will not tell you the algorithm that is used to calculate the TICK. So fundamentally, you can not say any of them is right or wrong. This lack of algorithm is the source of the final major difference between the different versions of the TICK, namely the initial value at the open.
All of the vendors, except Sierra, source the TICK only during regular market hours, and do not really tell you how they arrive at the initial value. What you can determine is it is not the initial up/down tick at the opening bell, as it regularly opens much higher or lower than the number of issues that actually traded in the first second. SO what are they using? Who knows because they will not tell you. This again becomes an issue for users because guru xyz says that when the ticks does blah-blah at the open.... The issue is they assume the version of the TICK that the guru us using is "correct".
Sierra is the only vendor that I am aware of that actually documents how they do the calculations, and provide full session (pre-regular-post) data. You have to turn it on in your chart, but you can see how the TICK runs in the pre-session and how it seamlessly transitions to the regular session. It is based on actual trades that go. It is sampled, but they do send thru all extremes.
At the end of the day, before you say any feed is right/wrong, ask if the algo is fully documented (including initial value at session start). Then you can compare apples to apples. Its also ok that they are different, as long as you learn the nuances of the particular feed and learn how to work with it.
The above also is true for the other internals.
The following 12 users say Thank You to aslan for this post:
Im just going to take a stab at a possible cause of this and say that SC gives you the number right from the opening bell when many stocks are not yet open while DTN (for example) waits a minute or so to publish their first print. Personally I have my A/D chart start 1 min after the bell because I find the first minute to be erratic as you have noted.
@aslan excellent post and i fully concur. are you using the SC data feed? Are you happy with it?
I am currently using it along side IQFeed, and will likely switch over shortly. Yes, it seems very good, and for the price it is pretty hard to beat. I do not concern myself with perceptions of the open values so could care less. Of course for others it might be very important.
For the initial value being so off, like I explained above, the issue is the algo. In the first second, there were not that many issues traded so I would tend to think the SC value is "correct". I suspect all of the other vendors use something else at the open to initialize their values (i.e. bid-ask mid being above below prior close, or pre-market trades, etc). The problem is they do not tell you what it is. It is days like today when you have extreme opens, that show that something else is used in the algo at the open. Even if you do not like SC values, they are documented and they are what they are.
The other interesting thing about the open, is after the first 10-30 min, the other vendors values come back to the sc values. Again, this shows they are using something else to initialize the values, and then as the issues actually trade, the values all come into alignment.
I would challenge anyone to get a real description of the full algo from any of the vendors. It won't happen.
I am having a bit of trouble processing this in my head. If SC starts to calculate the adv-dec values pre-market on a major gap down like this morning I would think that at the opening bell the data based on pre-market activity would be well below -216. Market makers and stop chasers will have already traded some kinda volume pre-market on a great deal of stocks to reflect a declining value from the prior days close.
Either way I agree with the idea that each is different and processing the info on a consistent baseline is one of the most important factors in using the internals to gain an edge.
Look at the data below. It is a 5 sec chart at the open today showing the net issues, adv issues, declining issues, and the traded issues per sec. At the open, the net was approx -215 and within the first 5-sec was at -1208. That actually seems pretty realistic to me. I suppose you could skip the first few seconds if you did not care about the 24-hour data (then you would open at ~ -1208). You can tweak the session for the symbol to start a few seconds (or a min) later to do that.
You do not see nearly as many pre-market issues traded as you might think (likely because only the big names are really traded much pre-market).
Please register on futures.io to view futures trading content such as post attachment(s), image(s), and screenshot(s).
The following user says Thank You to aslan for this post:
Here is a great example of the differences between platforms. one thing that kinda makes me think is at TS hung around the low for 3 one min candles but SC sampling every half second never got the new low reading.
Clearly you think TS is the gold standard, so why are you not using it?
If it is clearly correct, then where is the documented algorithm that they use? Do you even know what they are looking at to generate the data? Are they including all trades, all stocks, implied trades, are some things filtered out, is there any averaging, etc.? I am not implying their feed is wrong, just that they may not be measuring the same thing. The TS feed is considered one of the standards by many (I used to use it, but TS is such a dated platform I had to move on long long ago).
But what if the TS feed is wrong? How would you know? You can not call them on it because it is not defined. What if it is not measuring the same thing?
I am sure you are looking for the perfect divergence to trade, but guess what, today TS looks good, tomorrow BarChart will look better, and the next day SC will look better. You need to pick one of them and learn its nuances and run with it.
At the end of the day, Sierra is the only one that tells you what they are measuring and how it is done, which means it is the only one that can be verified to be correct if you had access to the source feed.