New engine released

[yuvaltaylor]

Thank you very much, Yuval.

First, please do run your experiment on each of the Legacy, Current w/prelim and on FactSet w/prelim engines to make sure you are generating the same list of stocks. If not, we may have stumbled on an issue, yes?

You wouldn’t get the same list of stocks unless you’re using a rather small universe because if you put WeeksIntoQ on its own line with no qualifications it’ll exclude stocks with WeeksIntoQ = 0 or NA, and the stocks with those values for WeeksIntoQ are going to differ between one server and another and between data providers.

Second, just to be clear… If stock ZZZZZ announces preliminary information on February 15th, 2020, files on May 1st, 2020, and S&P took a week in each case to process, its dates would be…

Press Release Date: February 15th, 2020
Effective Preliminary Date: February 22nd, 2020
Filing Date: May 1st, 2020
Effective Filing Date: May 8th, 2020

…correct?

Correct.

You wrote above that WeeksIntoQ used to be based on the Effective Preliminary Date. I don’t think that is so. I think it was based on the Effective Filing Date. Are you sure it was based on the Effective Preliminary Date?

Yes. Take ACU, for instance. It announced on April 17 and filed on May 8. WeeksIntoQ on Legacy is 3.

Third, I would discourage any subscriber from relying on LatestActualDays unless it turns out that FactSet is doing a much better job with this data point. Based on previous investigations, I’d say that this variable is theoretically perfect, but far from being so in practice. There are two issues, I believe. The first - and I admit I’m speculating based on my exploration - is that prior year values were not really point in time. The second is that analysts do NOT always update their EPS actuals on the date of the press release. They usually update their EPS actuals AS OF the date of the press release, but often at a later date. And sometimes they screw up and don’t update or their updates are not captured. This is especially so when analysts drop coverage.

Point taken.

Fourth, regarding FactSet, if the data we will have to work with is a) the press release date and b) the filing date, I strongly encourage p123 to create separate WeeksIntoQ variables based on each. WeeksIntoQ based on the press release date will allow subscribers to judge most accurately how early a company first presented information about a quarter. WeeksIntoQ based on the filing date will allow subscribers to judge where the company currently stands in relation to its last quarterly filing. DaysIntoQ would probably be better than WeeksIntoQ, by the way, since WeeksIntoQ can be derived from DaysIntoQ but not the other way around. If companies always released the same information on the press release and filing dates, perhaps the two data points would be redundant. But companies don’t. As mentioned before, I think it’s an unnecessary argument to debate which of the two dates is more useful. They each provide potential incremental value.

Couldn’t you use DaysSince(LatestFilingDate)?

[atw]

Yuval,

Thank you very much - once again - for your detailed reply.

I have some work to do on my end but this is very helpful and much appreciated.

Hugh

[paulnovell]

Finally had a good chunk of time this weekend to dig into differences in my Strategy Sims between Legacy Engine Compustat and Current Engine Factset databases. I’ll focus on one specific big difference for now since it affect a bunch of my strategies.

I have a risk management rule that is based on economic data and price performance of the index. Buy rule would be as follows:
Eval(Close(0,#UNEMP)>1.1*SMA(12,0,#UNEMP),Eval(BenchClose(0)>SMA(120,0,#Bench),1,0),1)

Sell rule would be similar but the opposite:
Eval(Close(0,#UNEMP)>1.1*SMA(12,0,#UNEMP),Eval(BenchClose(0)>SMA(120,0,#Bench),0,1),0)

These rules are causing large performance differences in strategies run on the Legacy engine with the Compustat data and the new engine with the Facset database. The results are even quite different if I use the Current Engine with the Compustat database. If I turn these rules off to test the fundamental part of the strategies, the results are close, not the same, but at least way way closer than with these simple rules turned on.

This is concerning to say the least.

Paul

[yuvaltaylor]

Finally had a good chunk of time this weekend to dig into differences in my Strategy Sims between Legacy Engine Compustat and Current Engine Factset databases. I’ll focus on one specific big difference for now since it affect a bunch of my strategies.

I have a risk management rule that is based on economic data and price performance of the index. Buy rule would be as follows:
Eval(Close(0,#UNEMP)>1.1*SMA(12,0,#UNEMP),Eval(BenchClose(0)>SMA(120,0,#Bench),1,0),1)

Sell rule would be similar but the opposite:
Eval(Close(0,#UNEMP)>1.1*SMA(12,0,#UNEMP),Eval(BenchClose(0)>SMA(120,0,#Bench),0,1),0)

These rules are causing large performance differences in strategies run on the Legacy engine with the Compustat data and the new engine with the Facset database. The results are even quite different if I use the Current Engine with the Compustat database. If I turn these rules off to test the fundamental part of the strategies, the results are close, not the same, but at least way way closer than with these simple rules turned on.

This is concerning to say the least.

Paul

Good catch. It looks like the buy rule gets activated on 4/6/09 on Legacy and 12/7/09 on Current, which explains the difference. We’re investigating and will report back soon.

[yuvaltaylor]

Good catch. It looks like the buy rule gets activated on 4/6/09 on Legacy and 12/7/09 on Current, which explains the difference. We’re investigating and will report back soon.

This will be fixed by tomorrow.

[paulnovell]

Thanks a bunch Yuval.

Paul

[paulnovell]

OK, on to the next issue I’ve noticed in my strategies.

A strategy that ranks stock by a custom shareholder yield formula produces substantially different (over 3% in terms of annual return) results on the Current Engine between Compustat and Factset. The custom formulas is:

((DivPaid(0,ann)+( EqPurch(0,ann)-EqIssued(0,ann)))/mktcap)*100

The custom formula is used in a custom ranking which is then used in the strategy. The strategy it is used in is “QEDGE: Large Stocks SHY Top 25 - 04”.

I tried to see if there was any big discrepancies in the individual stocks but didn’t really notice anything.

Paul

[atw]

What is the difference between FactSet Fundamentals and FactSet Coverage?

Also, will p123 be changing data providers for short interest information? If so, will that go live on the new engine prior to July 1st?

Thank you.

Hugh

[progscon]

Can someone restate what the settings mean for the below. I have gotten confused on what the different combinations mean and how to use them for running current rules as well as a backtest. thanks. David V

---

[rodbarcv2]

Since last weekend (which got the engine to use FactSet estimates), many of my sims that rely on sentiments are performing much worse with FactSet. For example, take “core: sentiment” and “basic: sentiment” ranks from portfolio123. The ranking graphs are different between Current Compustat and Current FactSet. (Current Compustat and Legacy Compustat look the same).

I didn’t have a chance to go through the factors / formula used in these ranking, is this something that can be investigated by p123 team?

[yuvaltaylor]

OK, on to the next issue I’ve noticed in my strategies.

A strategy that ranks stock by a custom shareholder yield formula produces substantially different (over 3% in terms of annual return) results on the Current Engine between Compustat and Factset. The custom formulas is:

((DivPaid(0,ann)+( EqPurch(0,ann)-EqIssued(0,ann)))/mktcap)*100

The custom formula is used in a custom ranking which is then used in the strategy. The strategy it is used in is “QEDGE: Large Stocks SHY Top 25 - 04”.

I tried to see if there was any big discrepancies in the individual stocks but didn’t really notice anything.

Paul

Paul, I’m afraid there’s not enough information here for me to investigate. Is that the only formula that chooses what stocks the simulation holds and what it buys and sells? Or are there other ranking systems, rules, etc. which might account for the difference? A number of the fundamentals are quite different between Compustat and FactSet.

[yuvaltaylor]

What is the difference between FactSet Fundamentals and FactSet Coverage?

Also, will p123 be changing data providers for short interest information? If so, will that go live on the new engine prior to July 1st?

Thank you.

Hugh

FactSet fundamentals are things like NetIncBXor and CapEx. FactSet coverage determines which stocks are in the All Fundamentals universe.

We will be changing data providers for short interest information but I don’t know if that will be before or after July 1st at this point, and I’m not sure whether or not users will be required to pay extra for that information. We’re working on getting answers to those questions.

[yuvaltaylor]

Can someone restate what the settings mean for the below. I have gotten confused on what the different combinations mean and how to use them for running current rules as well as a backtest. thanks. David V

See the first post here: https://www.portfolio123.com/mvnforum/viewthread_thread,12249

[yuvaltaylor]

Since last weekend (which got the engine to use FactSet estimates), many of my sims that rely on sentiments are performing much worse with FactSet. For example, take “core: sentiment” and “basic: sentiment” ranks from portfolio123. The ranking graphs are different between Current Compustat and Current FactSet. (Current Compustat and Legacy Compustat look the same).

I didn’t have a chance to go through the factors / formula used in these ranking, is this something that can be investigated by p123 team?

You should expect substantial differences. All the data providers count different analysts and have different methods of parsing those analysts’ numbers.

[rodbarcv2]

One of my models are not running at all, it uses Toronto Stock Exchange Universe, and it cannot find any stock during a 5 years simulation that meets this rule: EPSExclXor(0,qtr)>0 and EPSExclXor(1,qtr)>0 and EPSExclXor(2,qtr)>0 and EPSExclXor(3,qtr)>0 (last 4 quarters of earnings above breakeven). On Compustat it works fine (an on FactSet before this weekend change, it was working fine too).

Thoughts?

[WalterW]

Is there anyway to check in a log file whether Current-Compustat/Use or Current-FactSet/Use was chosen for a simulation? I’m see some rather large differences between the two data sources. That’s something new and I think it started recently - maybe today. But since I switch between the two often, it’s hard to confirm which results belong to each setting. Not having fun today.

Walter

[yuvaltaylor]

One of my models are not running at all, it uses Toronto Stock Exchange Universe, and it cannot find any stock during a 5 years simulation that meets this rule: EPSExclXor(0,qtr)>0 and EPSExclXor(1,qtr)>0 and EPSExclXor(2,qtr)>0 and EPSExclXor(3,qtr)>0 (last 4 quarters of earnings above breakeven). On Compustat it works fine (an on FactSet before this weekend change, it was working fine too).

Thoughts?

There’s a new bug and EPSExclXor is giving N/A no matter what quarter you choose. We’ll get this fixed ASAP.

[DenisD47]

Looks like DivPaidTTM is not working for Canadian REITs tonight in the screener using Factset.

Universe All Fundamentals - Canada
GICS(REIT)
DivPaidTTM > 0

Gives 2 REITS out of 39. Most DivPaidTTM are 0 or NA. This worked some days ago.

[aschiff]

Denis,
Many of the FactSet fundamentals were messed up around 5PM CT (7:50 hours ago). This was an unintended side-effect of work being done which will allow us to further address differences in FactSet fundamentals and make them behave more similarly. They should be back to normal by 4AM CT (in 3:10 hours). Thank you for your patience as we scramble to correctly implement all of the necessary features by the end of June.

[paulnovell]

Yep. I turned on and off all the other rules to isolate the performance difference to just that custom formula. I just ran it again and the Factset numbers got worse. Hopefully, this is because the 'FactSet fundamental are messes up…" right now…

Paul