User Roundtable: Chart Similarity Feature

tonyjimtrading: I’m an Australia-based trader and trade the US market overnight. I typically swing or position trade, but often trade around a position and am primarily an options trader. I usually trade part time, but get the opportunity to focus on it full time between consulting gigs.

Shan (e0): I’m a swing trader with relatively short holding periods — usually a few days to a few weeks for any given position. I trade the US equities market exclusively. On average, I aim to spend less than one hour a day on trading-related activities — scanning, prepping and managing orders.

tonyjimtrading: My most common practical use for the chart similarity feature is when I spot a breakout or big move that hadn’t been on my radar, signalling the possibility of a new theme.

I’ll then look for any significant price patterns in the lead-up to that breakout, like some prior move into consolidation, or a sharp move up or down that suggests the stock reacted to a news catalyst, and pay attention to the date range.

I then use similar charts to filter for other stocks that may have reacted in similar ways on similar time frames, so may be setting up for a similar move.

Shan (e0): Currently, when scanning, I exclusively use the new drawing variant of the similarity filter.

I use two similarity scans, both with the same basic criteria: ADR of at least 3% and a 20-day average dollar volume of at least $50 million.

The drawing for the first scan looks for flags with strong momentum — basically a strong upward move, followed by sideways consolidation. The second scan is similar, but with more of a pullback leading into a consolidation.

However, most of my trading revolves around hot themes, which I think has more edge than technical setups alone. This doesn’t mean I completely disregard charts — a lot of the time, the momentum of hot themes is reflected in the charts, so there’s good overlap. To look for stocks in the same theme, I also look at the similar charts from individual stock page [example].

Shan (e0): I run my two scans daily, usually looking through the top 60 results of each scan and adding the charts I like the most to a shortlist. Then, I try to identify if any of them belong to a specific theme. If so, I update my themed watchlist if necessary. If I don’t see an obvious theme, I look whether this could be a new theme and try to identify any related stocks.

The process of running a scan and compiling a shortlist only takes a couple of minutes for each scan. Updating themes, including looking for similar charts to a specific stock, usually takes between 5–30 minutes each day.

tonyjimtrading: The biggest benefit I see in the similar charts feature is that I mitigates the need to explicitly quantify the attributes of a chart that you’re looking for, and generate a scan matching those criteria.

Even for those with a coding background who can develop these things quickly, there’s a process of iterating through attributes and loosening criteria to see where results sets may best match what you’re attempting to find. This functionality just bypasses that entire process.

In terms of exploratory analysis, this feature truly facilitates and exponentially improves my ability — particularly in terms of time — to identify correlations between stocks that might sit outside the direct industry group.

Shan (e0): Running scans based on an ideal chart shape — in my case, stocks with momentum followed by consolidation — is a much more efficient process compared to scanning using traditional filters only and manually compiling the shortlist.

To be clear, this manual process is still needed, but I’m looking through dozens, as opposed to hundreds, of charts, since the similarity sorting has already placed all charts best matching my ideal pattern at the top.

It’s kind of like having an intern do 90% of the manual identifying process for you. If you were rating a pool of stocks in terms of technical setup, there would be lots of 1- and 2-star setups that you wouldn’t care about — they’re just noise.

This virtual intern selects only the 4- and 5-star setups for you to review.

tonyjimtrading: I’d love to have a set of pre-built favourites — so base chart and date range — just like how we have them for regular scans. Ditto for the pattern-drawing functionality.

An ability to look for inverse relationships would also be fantastic — things like trying to identify patterns in rotational markets: is money flowing into one theme, just as another theme is selling off? Being able to identify these sorts of lagging correlations is an area I haven’t seen any tool do well.

I also think that being able to look at particular date ranges, like when tariffs were introduced, would be super useful for performing back studies. Then we could see the biggest winners or losers at the time of a prior announcement, and do a similar charts scan across historical dates to see what else behaved in similar ways at those points in time.

This could help traders prepare and watch potential thematic groups ahead of time, rather than try to identify the theme after the move has already started.

Shan (e0): My wish list has two items.

First, I’d like a more meaningful representation of the similarity score.

Currently, it’s on a scale of 1–100. However, since the similarity is calculated mathematically, even quite dissimilar stocks would still score around 60. Maybe a grading system of A–F, where A would be 95–100, for example, would give a better representation of the similarity of two stocks relative to every other stock in the market.

Second, the mathematical model of comparing charts works great for chart patterns but, on its own, doesn’t say anything about how similar two companies are in terms of the nature of their business.

If we can find a way to match companies fundamentally, in terms of what they do as a business, working on theme lists would be much easier.