Correct pool model & decent statistical predictions for the next draws

[PMM]

Hi

It's quite a stretch to answer that, because the model is iterative so it gets more and more inaccurate with every subsequent prediction. But based on the limited data we have so far, I would say: 350 rather not possible this year, 380 quite possible in August or September. Remember that the difference between 350 and 380 is very very huge.

1. Don't forget that applicants should be starting to get approved LMIA to support PR, the ones where the employer doesn't have to pay, this may boost the score required in the coming months.

 

[Michau]

I will be watching the spreadsheet to adjust to such situation properly

 

[Karolis]

I took some time and ported the algorithm to R and uploaded it on github
https://github.com/karolisjakaitis/ee_draws

Future plans that are possible:
1) Make script automatically read spreadsheet (currently it uses march 29 export)
2) Make R Shiny dashboard that would include some visualizations, model output, etc.

You are free to play with my code and if you make any changes just make a pull request. Also, if you see any errors let me know. Something might have slipped.

Cheers!

 

[rubyracer]

Just to add, only percentage representation is not import. The distribution has to match too , for getting accurate predictions

 

[rojiv94]

OK, so now we are after 6 draws, we can create a decent statistical model of the pool based on the Express Entry spreadsheet.

We know (just like I predicted here: http://www.canadavisa.com/canada-immigration-discussion-board/a-possibly-plausible-theory-about-the-number-of-people-in-the-pool-and-scores-t271152.0.html) that the first draws had unusual number of 600+ applicants due to errors, and this is now confirmed by many people who got 600 points added by CIC's mistake. So the results of first draws are really irrelevant and we should not look at them.

The last two draws give us much better information, and we can see from them that the current EE spreadsheet contains about 4-5% of the pool, just like the previous year's spreadsheet for FSW (nobody should be really surprised by that).

So let's build a statistical model based on the current Excel contents. We need the following parameters:

* The relation of the spreadsheet population to the real pool population, let's assume it being from 4.0% to 5.5%.

* The distribution of people scores. Let's take the list of all scores from the spreadsheet below 600 and it will give us a good score distribution between applicants; not a perfect one, but definitely good enough. Then there is also this +600 factor, we will take it into account by introducing a probability of someone receiving LMIA or PNP. In the spreadsheet, 48 people out of 974 have above 600 points so we assume that this probability of someone having LMIA/PNP is around 5%.

* The number of people in the pool on day 0 (which is today). The spreadsheet contains today 813 people without ITA (that is, below 453 points).

* The number of people being added to the spreadsheet daily. From March 1 to March 28 there were 174 new entries in the spreadsheet, this gives on average 6.2 new entries per day.

* The dates and counts of future draws. Nobody knows these of course, but let's assume 2 draws per month, 1600 people each - this looks like a good assumption based on the draws done in March, and also based on immigration targets for 2015.

That's all the assumptions we need. They are not very accurate yet, but at this point should be accurate enough to build a simple model. So this is the actual model parametrized (sorry it's a PHP code - I'm a Web designer ):

http://pastebin.com/wEHvGwyB

What does the model tell us? Based on the assumed relation of the spreadsheet population to the real pool, we get the following results:

If the spreadsheet represents 5.5% of the pool:

April 1st draw: cutoff score = 437
April 2nd draw: cutoff score = 427
May 1st draw: cutoff score = 415
May 2nd draw: cutoff score = 409
June 1st draw: cutoff score = 400
June 2nd draw: cutoff score = 394
July 1st draw: cutoff score = 388
July 2nd draw: cutoff score = 384

If the spreadsheet represents 5.0% of the pool:

April 1st draw: cutoff score = 438
April 2nd draw: cutoff score = 429
May 1st draw: cutoff score = 419
May 2nd draw: cutoff score = 413
June 1st draw: cutoff score = 408
June 2nd draw: cutoff score = 401
July 1st draw: cutoff score = 396
July 2nd draw: cutoff score = 392

If the spreadsheet represents 4.5% of the pool:

April 1st draw: cutoff score = 439
April 2nd draw: cutoff score = 432
May 1st draw: cutoff score = 423
May 2nd draw: cutoff score = 417
June 1st draw: cutoff score = 412
June 2nd draw: cutoff score = 409
July 1st draw: cutoff score = 403
July 2nd draw: cutoff score = 399

If the spreadsheet represents 4.0% of the pool:

April 1st draw: cutoff score = 441
April 2nd draw: cutoff score = 435
May 1st draw: cutoff score = 428
May 2nd draw: cutoff score = 421
June 1st draw: cutoff score = 416
June 2nd draw: cutoff score = 413
July 1st draw: cutoff score = 411
July 2nd draw: cutoff score = 407

Personally I am leaning towards 4.5% - 5.0%. So that means that we should reach 400 around June/July.

Another interesting value is the final value when the draw score stabilizes. This is highly dependent on the ratio of new people added to the pool, to the number of people drawn, and this first parameter is derived from the spreadsheet and also not very accurate. But, just to give you a ballpark number, with the 5.0% spreadsheet-to-pool assumption, after two years (!) the model reaches somewhere around 360, and with the 4.5% assumption it reaches about 365. The actual stabilized score may be lower if the number of new people adding themselves to the pool will be dropping, or the number of people drawn will be higher than 3200-3300 per month as we see now. But generally I would say looking at these first approximations, that people above 360 should finally get their ITAs, though the wait may be really long.

Totally wrong assumptions. Very poor. -1 for you
Sorry, but I must say, all persons follows this thread are pessimists.

 

[preeti.msingh109]

Totally wrong assumptions. Very poor. -1 for you

Lol..because Rojiv falls nowhere !!

 

[fpnc2983]

Totally wrong assumptions. Very poor. -1 for you

can you elaborate why he is wrong?

 

[rojiv94]

can you elaborate why he is wrong?

Spreadsheet represents to 10% of EE pool. Now (at 453 cutoff score) it is 5.01%, but it is sharply increasing and when the cutoff score comes below 400, I am sure that it will reach 10% and reflects a true picture.

I heard that, after sixth draw all pesimists had ran away. But it seems they are back with a new form of prediction

 

[fpnc2983]

Spreadsheet represents 10%. Now it is 5%, but it is gradually increasing and when the cutoff score comes to 400, I am sure that it will reach 10% and reflects a true picture.

I heard that, after sixth draw all pesimists had ran away. But it seems they are back with a new statistical prediction

what's your basis that it represents 10% of the pool and not 5?

don't get me wrong but
this statistical model looks pretty optimistic to me.
the predictions are quite conservative. not too low and not too high.
in the past pessimists predicted that score will never drop below 500 since they said CIC will maintain a high cutoff score.

 

[hamoonExpressEntry]

Good job! A brilliant analysis!

 

[Karolis]

Spreadsheet represents to 10% of EE pool. Now (at 453 cutoff score) it is 5.01%, but it is sharply increasing and when the cutoff score comes below 400, I am sure that it will reach 10% and reflects a true picture.

I heard that, after sixth draw all pesimists had ran away. But it seems they are back with a new form of prediction

How can you say that it is sharply increasing when you have only 2 points of data.... e.g. increase from 0 to 10 is not a sharp increase if the next points are 10.5, 10.51 and 10.53.
Lets just wait and see how accurate the model is at the next draw. it is predicted that if the next draw has 1600ITA then crs will be around 438-441.

 

[gmduncanidaho]

I spoke with an immigration lawyer today and he said I should expect an ITA in September or October with my current score (431)

 

[preeti.msingh109]

I spoke with an immigration lawyer today and he said I should expect an ITA in September or October with my current score (431)

he might not have come across previous draw's cut off!!! dnt believe him...u ll get it much before

 

[Karolis]

I spoke with an immigration lawyer today and he said I should expect an ITA in September or October with my current score (431)

I think you should get in the next 2 draws. He told you the safest possible timeline so he is not wrong

 

[gmduncanidaho]

I think you should get in the next 2 draws. He told you the safest possible timeline so he is not wrong

he might not have come across previous draw's cut off!!! dnt believe him...u ll get it much before

Thanks I was thinking the same, expecting an invitation end of April or May for sure (Score is going down on April 15 also my work permit is expiring the same day -worst birthday ever). I explained this to him as well.

He sounded like he had gotten the last draw's information. He said he doesn't expect the score to go lower than 450, they are only inviting LMIAs and "geniuses". He said, and I'm quoting his exact words here, that "the new system is bull****" and when they invite people with 750 points, they are inviting "****" (the previous censored word without the bull part) and leaving people like me out.
He still said I hope you are right and I'm wrong, but he sounded pretty sure that a drop won't happen anytime soon. I guess we'll see what happens end of April. It all depends on the amount of LMIAs and PNPs issued within the next few weeks.