HGA grading

[miscus555]

Originally Posted by growdoe

Since you brought up SGC.... SGC is currently exceeding expectations. I sent 40 cards, two subs, at the end of January, and they are in my hands now. Many on these boards have commented that they are now getting 15/20-day turnarounds, when they expected 60-70. These are facts.

If you want to apologize to the hobby for how you handled getting swamped and talking smack, you do exactly what SGC is doing now. I certainly appreciate their efforts and easily see how they are competing in the game.

That great for them. Didn’t know that. They look to be the option if you plan on never selling your cards though. I’ll tell you this, SGC has been said to be the best actual graders out of all of them but the public never honored them in the secondary market since they don’t bring in the returns. If I was a collector looking for a deal, I’d be buying SGC cards left and right and resubbing them to PSA since their grading of a card is very good

[growdoe]

Originally Posted by miscus555

I’ll tell you this, SGC has been said to be the best actual graders out of all of them

I don't want to get too far off-topic from this HGA thread, but if we still talking SGC...

What you say does align with conventional wisdom that a "10" from SGC, on average, pound for pound, is a stronger Gem Mint grade. So let's carry this line-of-reasoning a little further. Do you think future investors will consider, especially when after 5, 10, 15, 20 years go by and the hobby is constantly reminded each year that SGC is better at grading, that investing in a stronger 10 from a company with the reputation for being the best actual graders out of all of them might be something worthwhile? We'll see. Things do change.

[rfgilles]

Originally Posted by growdoe

I don't want to get too far off-topic from this HGA thread, but if we still talking SGC...

What you say does align with conventional wisdom that a "10" from SGC, on average, pound for pound, is a stronger Gem Mint grade. So let's carry this line-of-reasoning a little further. Do you think future investors will consider, especially when after 5, 10, 15, 20 years go by and the hobby is constantly reminded each year that SGC is better at grading, that investing in a stronger 10 from a company with the reputation for being the best actual graders out of all of them might be something worthwhile? We'll see. Things do change.

Grading better be completely automated in 5 years.

[growdoe]

Originally Posted by rfgilles

Grading better be completely automated in 5 years.

A human will always verify, unless the buying part gets automated too.

[The_Reverend]

Originally Posted by rfgilles

HGA has only been on the market for a few months and with capped submissions. They honestly don't need people to crossover cards from PSA, as long as their is continued interest to submit ultra modern, raw cards. Some of the ultra modern cards could end up being "tomorrow's big dogs."

If HGA starts providing scans with reports of the card flaws detected by automated grading, as Tyler said they will, their grading will be more credible than the other TPGs. In the future maybe a PSA 10 subjectively graded by a human will be less desired than an objectively graded HGA card.

You have to start somewhere with automated grading, and I'm excited to see where HGA goes over the next year.

As I said, I wish them luck, but for me ROI matters when it comes to grading. The wait is a pain and I could lose some money, but I have a strategy that puts cards in my hand at the right time. I use express for PSA TO GET them faster. I will wait and see. Most businesses sink or swim in 18 months so we shall see. There will be those who buy the slab and not necessarily the card, but fir me I am sticking with PSA.

[growdoe]

I haven't seen this talked about much, so adding here...

Each card company will have their own flavor of AI, and each one will behave differently because each one will be trained differently. All the bias/expertise of PSA, or SGC, or HGA, will get encoded into their respective AIs. So, if one company now has the reputation for being the "best graders" currently, they will likely still be the best graders with AI because their algorithm was given the "best" examples to work from.

You take an algorithim, let's say a Recurrent Neural Network. That's just one type of algorithm. There are hundreds of different types. Each grading company will pick the best one they think will be easiest to maintain. There will be no common AI system that they all use, because each company will think they can improve it themselves.

You then feed the algorithm Examples. You show the AI computer vision a card, and then you tell it what grade that card is. You do this tens of thousands of times to build the model. Some algorithms need fewer training examples, some need more.

The net result will most likely be that the grading companies will save money from operational cost, and the grading outcomes will remain the same from our perspective. Those who shine now at grading, will most likely continue to shine with AI grading.

And we'll stil be arguing on who is the best graders.

[rfgilles]

Originally Posted by growdoe

I haven't seen this talked about much, so adding here...

Each card company will have their own flavor of AI, and each one will behave differently because each one will be trained differently. All the bias/expertise of PSA, or SGC, or HGA, will get encoded into their respective AIs. So, if one company now has the reputation for being the "best graders" currently, they will likely still be the best graders with AI because their algorithm was given the "best" examples to work from.

You take an algorithim, let's say a Recurrent Neural Network. That's just one type of algorithm. There are hundreds of different types. Each grading company will pick the best one they think will be easiest to maintain. There will be no common AI system that they all use, because each company will think they can improve it themselves.

You then feed the algorithm Examples. You show the AI computer vision a card, and then you tell it what grade that card is. You do this tens of thousands of times to build the model. Some algorithms need fewer training examples, some need more.

The net result will most likely be that the grading companies will save money from operational cost, and the grading outcomes will remain the same from our perspective. Those who shine now at grading, will most likely continue to shine with AI grading.

And we'll stil be arguing on who is the best graders.

Thanks for your insight. Though if companies are transparent with their grading standards, and provide reports of the card imperfections, like HGA says they will provide in the future, it will be easier to evaluate who is the 'best grader'. I would also think that although AI grading may differ slightly between companies, there will be greater consistency for each company.

[HalfNipponese]

Originally Posted by growdoe

I haven't seen this talked about much, so adding here...

Each card company will have their own flavor of AI, and each one will behave differently because each one will be trained differently. All the bias/expertise of PSA, or SGC, or HGA, will get encoded into their respective AIs. So, if one company now has the reputation for being the "best graders" currently, they will likely still be the best graders with AI because their algorithm was given the "best" examples to work from.

You take an algorithim, let's say a Recurrent Neural Network. That's just one type of algorithm. There are hundreds of different types. Each grading company will pick the best one they think will be easiest to maintain. There will be no common AI system that they all use, because each company will think they can improve it themselves.

You then feed the algorithm Examples. You show the AI computer vision a card, and then you tell it what grade that card is. You do this tens of thousands of times to build the model. Some algorithms need fewer training examples, some need more.

The net result will most likely be that the grading companies will save money from operational cost, and the grading outcomes will remain the same from our perspective. Those who shine now at grading, will most likely continue to shine with AI grading.

And we'll stil be arguing on who is the best graders.

You are probably right and it won't really matter, but what will happen is competition. That'll drive improvements in customer service and innovation. The industry hasn't evolved in over 20 years and I hope the grading companies will appreciate their customers more.

[Sampson10k]

SO THEY LOST MY ORDER. IT WAS DELIVERED LAST FRIDAY. NO IDEA WHERE IT IS. CUSTOMER SERVICE IS NON EXISITENT. SO FAR IM REALLY DISAPPOINTED IN MY 400 DOLLARS IVE ALREADY BEEN CHARGED

[mooselang]

Yikes. That is not a good look this early in the game. I think I have heard that they are more responsive on their Facebook page. Please keep us updated on what happens!

[CheckEmOut]

Originally Posted by Sampson10k

SO THEY LOST MY ORDER. IT WAS DELIVERED LAST FRIDAY. NO IDEA WHERE IT IS. CUSTOMER SERVICE IS NON EXISITENT. SO FAR IM REALLY DISAPPOINTED IN MY 400 DOLLARS IVE ALREADY BEEN CHARGED

What company are you taking about? I reached out to HGA on their website chat and had a response within 2 hours and they immediately took care of my issue (nothing major - just wanted to change the slab color scheme).

[salthill]

Originally Posted by Sampson10k

SO THEY LOST MY ORDER. IT WAS DELIVERED LAST FRIDAY. NO IDEA WHERE IT IS. CUSTOMER SERVICE IS NON EXISITENT. SO FAR IM REALLY DISAPPOINTED IN MY 400 DOLLARS IVE ALREADY BEEN CHARGED

Bummer of a first post.

[salthill]

Originally Posted by growdoe

You then feed the algorithm Examples. You show the AI computer vision a card, and then you tell it what grade that card is. You do this tens of thousands of times to build the model. Some algorithms need fewer training examples, some need more.

Your whole post is spot on.
And I’m going to guess that the training data HGA used included PSA and BGS graded cards. Otherwise it’s a Herculean effort to generate enough examples to train it.

[rfgilles]

Originally Posted by salthill

Your whole post is spot on.
And I’m going to guess that the training data HGA used included PSA and BGS graded cards. Otherwise it’s a Herculean effort to generate enough examples to train it.

Yeah they used both PSA and BGS graded cards.

[Sampson10k]

Originally Posted by salthill

Bummer of a first post.

Tell about it dude. Im frustrated to say the least. At least tell me you have my cards.

[Sampson10k]

Originally Posted by CheckEmOut

What company are you taking about? I reached out to HGA on their website chat and had a response within 2 hours and they immediately took care of my issue (nothing major - just wanted to change the slab color scheme).

Yeah i sent a message at 11 today. Nothing. At 4 sent a reponse. Almost within seconds. They tell me it takes 24-48 hours to show. I said great it got there last friday. Sent order number and tracking. Nothing for an hour. I finally say hello? And they say we will have to get back to you. Lilke really?

[Sunsfan86]

I just refreshed for an hour waiting for it to open and was assigned spot 1,800.....with only 1500 card openings. Wow.

[Bcr]

This whole lottery thing is bogus. I'm better off sending to PSA at this point and waiting instead of having no chance at all.

[mooselang]

I’m not submitting until next week or the week after, but I wanted to check it out just to see how it worked.......and to witness the shitshow


I was 1,508. Now in the 800s with 35 estimated mins left

[Mamschler1]

Well this is fun.... im sitting in position #2..... and they claim to have technical difficulties and paused the line.... I figured they would sell out by the time my spot hit (164) but teetering on the ledge of success is just mean.

[bradley9999]

I had 2 browsers open, one was spot 13 when it opened, and got the error, the second one was spot 2 when they paused. As soon as they opened it back up I got in and subbed 2 orders. Cant wait to see what they look like.

[Sunsfan86]

Originally Posted by bradley9999

I had 2 browsers open, one was spot 13 when it opened, and got the error, the second one was spot 2 when they paused. As soon as they opened it back up I got in and subbed 2 orders. Cant wait to see what they look like.

I don't get how it works. I filled it out this morning and had the page open for hours. Why did I get 1800?

[bradley9999]

Originally Posted by Sunsfan86

I don't get how it works. I filled it out this morning and had the page open for hours. Why did I get 1800?

It doesn’t matter how long you have it open. Once the queue starts everyone gets a random spot.

[PostBaloney]

managed to get in tonight! 20 card order ranging from 2020 prizm draft to a 90's ray allen net asset die cut. Sent them all kinds of different things to see how well they turn out

[HalfNipponese]

I was listening to a Podcast in the car and I thought I heard the CEO say they are thinking about a 20 card limit. I think that would be a good idea so more people can get in.