I had my CV reviewed professionally a couple of days ago and was amazed to learn that most recruitment firms and HR departments do not read CV’s when they get them. In response to the huge numbers of CV’s they receive they use software to scan them first; to extract what they consider to be the key relevant information and look for key words. The software then ‘grades’ the CV as being suitable for the position or not. If you score high enough you get considered and then a person may read your CV.
That made me think I need two CV’s; one for the machine and one for a person to read. I was prompted by a post on LinkedIn by Goncalo Nunes who said he had replied to 100 job adverts and hadn’t heard back from most of them.
LinkedIn says it is about connecting people but in this case it seems it is not.
It could help people looking for jobs by requiring those who advertise positions on its platform to stipulate whether the CV submitted would be scanned by a software programme or read by a person. I guess that if the recruiter said which programme they were using you could alter your CV to add words or phrases; if you knew which ones the programme favoured. All of which to me seems to defeat the point of the CV which I take to be a personal statement rather than an attempt at box ticking.
Progress has shown me to be wrong. It order for my CV to be considered I need to put in more key words and leave out some of the explanation.
That suggests that a lot of good candidates are not being considered because the software does not like the words they use in their CV. Whilst many people adapt their CV’s to a specific company or role if the base words don’t meet the software required target words they are being excluded.
Also it would appear from Mr Nunes experience they don’t even get a response saying why they were not being considered. A human reading a CV is better at understanding different peoples style of words. A human can also give feedback to the candidate so the person can either adapt their CV or change their job search parameters to better meet their aspirations.
Advertisers could also be required to give feedback to applicants. That would really sort out the good recruiters who really care about the candidates as well as providing their clients with the best applicants. It's also worth noting that sometimes the best candidates are not those who tick all the boxes but the ‘left field’ ones who can really add a new skill or angle for the client.
Good recruiters are the ones that take the time to understand both what their client needs and what candidates can offer. Recruiters don’t get clients by scanning the client's website and seeing if it has enough key words but by meeting and talking to them. It would be nice if they extended the same courtesy to the candidates.
That then got me thinking about broker research, and an article in this week’s FT Macro hedge funds enjoy unlikely renaissance (20 August) about how some hedge fund managers who had exited the market a few years ago due to the lack of volatility are coming back as their ‘old school style’ is making money.
It noted in ’the old days’ portfolio managers could take advantage of 'sparser and slower financial information, riskier economic policymaking and less efficient markets’ something that is not coming back says the article.
Actually to some extent it is coming back; as some firms employ more AI to preform quicker analysis of more data. Some funds will have a technology edge, even if it's only seconds that can make a difference. But that makes them like recruitment companies. Scanning research and the internet for key words to determine how useful a piece of research is. Or deciding what sentiment is like based on phrases in the media or what sectors are in momentum based on the occurrence to key words in the press or search engines. All that is dependent on the search criteria put into algorithms to start with.
The FT article notes that
“All the best macro hedge funds were great aggregators of information. They then come up with a non-consensus thesis, put on a position and become great storytellers. And then a huge wave of money would follow them,” But that too has changed with more money going into passive or systematic strategies and an increase of money to the quant funds.
All of which is progress but does it add value?
I started in broking as an analyst and so I like fundamentals but I have also embraced technical's and the ever increasing sources of data. As a trader when I started we used a machine to physically time stamp tickets, today the time stamp is electronically noted when the order arrives in the FIX system. 18 years ago I was on the Hong Kong FIX committee when decisions were being made about what information should go into the protocol, look how far things have changed since then. I have seen many useful and enhancing changes. But I worry that some of applications of AI are actually undermining the business.
Good analysis and Fund mangers build up thesis, they question management and query the managements numbers and business guidance. They do their own research and that makes them standout from those that just take the company’s numbers and put them in their model.
A good analyst probably understands the company’s business as well as, if not better than the management. But there seem to be only a few of them out there; as evidenced by the fact that each quarter in the US we see a large proportion of companies ‘outperform’ analysts forecasts. That suggests things are not right. It raises the question that maybe analysts are relying too much on the data generated by AI and not enough on ‘old fashioned’ questioning and research?
It reminds me of the part in the film Top Gun, when the candidates, newly arrived, are being shown a film about the Korean War and told
'During Korea, the Navy kill ratio was 12-to-1. We shot down 12 of their jets for every one of ours. During Vietnam, that ratio fell to 3-to-1.
Our pilots become dependent upon missiles. They had lost some of their dogfighting skills.
Now Top Gun was created to teach A.C.M. Air Combat Maneuvering. Dog fighting.’
Maybe the same is true of brokering. The fact that so much time and effort is going into automation, scanning, Big data and AI; it begs the question; have we lost some of our basic skills and become too reliant on data. If the algorithm is slightly wrong or data isn’t aligned with whatever the programme is looking for, then the results will give a false impression. As with all technology it can be useful but an over reliance on it can lead to disaster.
It is also true in trading, reliance on algorithm’s in trading. When they were first introduced there are cases of algorithms competing with each other in the same stock; leading to some very poor executions and strange stock price movements. They have become much better with refinement and more historical data but a good trader can still spot them and take advantage of them, call it the Dog fighting of trading vs missiles. Good traders and sales traders will alert the client and present him with the options of how to best use the situation.
A lot of hedge funds have made money this year because they didn’t go with the trend and didn’t believe that the world was ending in March. Sometimes the non-consensus theory combined with plain common sense is better than the leading AI.
By the way if you want your CV checked, go to Glassdoor; when you sign up they offer a free CV appraisal; see how you score.