MarketsFlow is returning to Crowdcube

(Kenny Grant) #1

I invested a small amount in this lot last time they raised (last year), and thought people here might be interested in it.

I find the idea quite interesting - algorithmic trading is popular with professionals but has not been offered to retail investors, and this takes the novel approach of using machine learning applied to markets, which should in theory be more adaptable to market conditions than a simple algorithm concocted by humans, with the downside that its failure modes may not always be apparent. I’ll be interested to see where this goes.

6 Likes

(Kenny Grant) #2

This is now open for investment for everyone:

1 Like

#3

I don’t have access to the restricted docs yet so it is hard to tell but is Marketflow’s promise this?: “our product gives the investor an edge on the market because its algos are super clever”

CEO looks like he’s hiding in the photos!

0 Likes

#4

I invested a small amount in their last round, it’s gone well, a 10x-ish paper gain.

The biggest surprise was how forth coming Tom Nash was with updates, so all along the journey investors were notified of how Marketsflow was progressing, which was great.

Marketsflow has progressed a lot since the last round, from acquiring the relevant licenses to testing out their model a bit more.

I’m also a bit skeptical about their performance, 25.55% seems to good to be true, but then again it’s only 1-year, and it depends on the portfolio sizes etc, as things that work on smaller portfolios don’t necessarily work on larger ones.

It’s great to see they are reducing minimum investment amounts to £10, plus the app, which indicates they are steering away from the current look/feel/initial impression you get from the website.

Would be good if Freetrade could swing a way to get Marketsflow to do a QA?

2 Likes

#5

Interesting area. I am quite sceptical that quants will perform well in the future. Having seen the amount of investment gone into this space and lack of returns for the risk. (My personal view)

In saying that. Some will do very very well and there is always going to be a place in the market for alternative providers.

I’m curious how alternative data like twitter trends and google searches will impact trading going forward.

0 Likes

(Kenny Grant) #6

Yes I honestly have no idea how this one will work out, or whether their returns are sustainable and based on a significant advantage, or how reliable their figures are. I’ve invested relatively small amounts in these two rounds, and am going to try out the investment product in a small way this year, I’ll report back if it seems interesting when they launch their app and I try it out. I certainly would not put my life savings into this at this stage.

From their released info they seem to be using a brokerage account the client opens in their own name, and using machine learning trained on historical data (I hope a lot of it!) to do momentum investing for clients on a large selection of relatively safe large cap stocks (to minimise downside risk presumably). They are also considering an ETF which would be a useful indicator of their real returns using this strategy and probably a safer way to get exposure to this sort of thing.

There are obviously people making money with active trading strategies like momentum investing and simple or complex hand-written algorithms, but I haven’t heard of many using machine learning (a hard claim to verify, so I’m skeptical there). This lot do seem data driven (not chartists or working on sentiment), and if they are using significant machine learning that could be a game-changer here, because it takes the quants out of the picture to a great extent - the system learns what works and what doesn’t, it can be trained on vast amounts of historical data, and in theory it will do better than humans at momentum investing.

As far as I know they don’t use external data like twitter trends - I’d be hesitant to use that sort of data as it is so easily gamed and often wrong.

2 Likes

(Kenny Grant) #7

Yes, you are right to be skeptical!

The reason I feel they are more interesting than most proponents of this idea is they are proposing using machine learning to do it, which has had huge advances in competitive game play (deep mind, openai), in autonomous vehicles (waymo, tesla) and in image recognition (google) ) - that’s a step change from traditional approaches which as far as I know use algorithms designed by humans which of course are liable to behaving badly in unexpected conditions.

We’ll see, I find the idea interesting and suspect bots will replace human traders entirely in the long term, but this lot may well be not those who gain from that shift.

1 Like

#8

I might be too skeptical and completely mistaken, but I assume that the majority of all trading is already done by MLed algos (all that high frequency trading stuff), and that every institutional asset mgr has been deeply into ML for ages. But I don’t know.

I suppose my other thought is: if I had the ability to consistently make 20-25% annual return investing with small drawdowns, then I’m not sure I’d even bother selling a software product for 30/month? Maybe I’d just do the 1&20 managed portfolios. Or I’d not even bother with that - I’d just invest my own money and it’d grow quickly enough, and then I wouldn’t need to bother with all that running a company, regulatory and compliance, and crowdfunding stuff. Dunno.

It smells a bit over-promised but I will read the docs with clear eyes and an open mind. (Maybe I’m skeptical because at some level I’m annoyed I didn’t invest in the previous round haha.)

1 Like

(Kenny Grant) #9

It’s not really required for HFT, but I don’t know either I’m afraid how prevalent it is. I do know people I’ve talked to about momentum investing specifically didn’t mention it a few years ago, but I’m sure some people must be using it. Would be really interested to hear from anyone with more experience in the industry how prevalent it is.

:rofl: I’m not sure the returns are quite that good! I think they plan to sell to some larger company to cash out but also charge significant fees.

I’d like to actually try their managed portfolio with a small amount of money (< £1000) for a year or two to see what they do, so am going to see if I can do that. I’d also like to see them get approval to hold client money (as opposed to manage it) from the regulators, and think I’d rather something safer for the majority of my investments.

Any returns on the crowdcube shares are definitely paper returns at this early stage!

0 Likes

#10

20-25% is their claim…

My guess about HFT was that if you’ll spend to get microwave line-of-sight stuff to get the tiniest of speed edges then you’ll look carefully at any tech that promises any edge, incl ML. :man_shrugging:

0 Likes

#11

Momentum trading has been around for a long time. I’m guessing ~20 years. It was extremely profitable pre financial crisis.

Ever since the financial crisis quant trading hasn’t done so well.

Firms are using ML - or are trying. Talking to people in the business they all say it is much more difficult than people make out. There is no easy win and there are so many people trying that spreads become tight and then you require leverage to magnify the gain (or loss)…

I too am extremely sceptical of their offering. Anyone advertising a return that high has something to hide or they are extremely lucky.

0 Likes

(Morty) #12

It sounds interesting, but that’s a very high evaluation given their current revenue. Plus, unless I’m missing something, their forecasts are based on that revenue increasing by almost 100x times over the course of the next 12 months.

0 Likes

(Kenny Grant) #13

Yes the forecasts are insanely optimistic I feel and I’m skeptical of their valuation (like most at this stage, it’s not based on much).

0 Likes

(Morty) #14

Agreed - looking at numbers, it seems to be based on the arrival of a £10 minimum investment and a 10 minute sign up, but there’s no transparency around when those features are due to arrive.

Plus my thinking is that if they’re not here now, then the timeline is shorter: that’s less than 12 months to get the features up and running, market them and then see the results.

It feels like quite a push.

0 Likes